1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, 56 FAULT_WRITE_IO, 57 FAULT_MAX, 58 }; 59 60 #ifdef CONFIG_F2FS_FAULT_INJECTION 61 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 62 63 struct f2fs_fault_info { 64 atomic_t inject_ops; 65 unsigned int inject_rate; 66 unsigned int inject_type; 67 }; 68 69 extern const char *f2fs_fault_name[FAULT_MAX]; 70 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 71 #endif 72 73 /* 74 * For mount options 75 */ 76 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 77 #define F2FS_MOUNT_DISCARD 0x00000004 78 #define F2FS_MOUNT_NOHEAP 0x00000008 79 #define F2FS_MOUNT_XATTR_USER 0x00000010 80 #define F2FS_MOUNT_POSIX_ACL 0x00000020 81 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 82 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 83 #define F2FS_MOUNT_INLINE_DATA 0x00000100 84 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 85 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 86 #define F2FS_MOUNT_NOBARRIER 0x00000800 87 #define F2FS_MOUNT_FASTBOOT 0x00001000 88 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 89 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 90 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 91 #define F2FS_MOUNT_USRQUOTA 0x00080000 92 #define F2FS_MOUNT_GRPQUOTA 0x00100000 93 #define F2FS_MOUNT_PRJQUOTA 0x00200000 94 #define F2FS_MOUNT_QUOTA 0x00400000 95 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 96 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 97 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 98 #define F2FS_MOUNT_NORECOVERY 0x04000000 99 #define F2FS_MOUNT_ATGC 0x08000000 100 101 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 102 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 103 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 104 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 105 106 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 107 typecheck(unsigned long long, b) && \ 108 ((long long)((a) - (b)) > 0)) 109 110 typedef u32 block_t; /* 111 * should not change u32, since it is the on-disk block 112 * address format, __le32. 113 */ 114 typedef u32 nid_t; 115 116 #define COMPRESS_EXT_NUM 16 117 118 struct f2fs_mount_info { 119 unsigned int opt; 120 int write_io_size_bits; /* Write IO size bits */ 121 block_t root_reserved_blocks; /* root reserved blocks */ 122 kuid_t s_resuid; /* reserved blocks for uid */ 123 kgid_t s_resgid; /* reserved blocks for gid */ 124 int active_logs; /* # of active logs */ 125 int inline_xattr_size; /* inline xattr size */ 126 #ifdef CONFIG_F2FS_FAULT_INJECTION 127 struct f2fs_fault_info fault_info; /* For fault injection */ 128 #endif 129 #ifdef CONFIG_QUOTA 130 /* Names of quota files with journalled quota */ 131 char *s_qf_names[MAXQUOTAS]; 132 int s_jquota_fmt; /* Format of quota to use */ 133 #endif 134 /* For which write hints are passed down to block layer */ 135 int whint_mode; 136 int alloc_mode; /* segment allocation policy */ 137 int fsync_mode; /* fsync policy */ 138 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 139 int bggc_mode; /* bggc mode: off, on or sync */ 140 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 141 block_t unusable_cap_perc; /* percentage for cap */ 142 block_t unusable_cap; /* Amount of space allowed to be 143 * unusable when disabling checkpoint 144 */ 145 146 /* For compression */ 147 unsigned char compress_algorithm; /* algorithm type */ 148 unsigned char compress_log_size; /* cluster log size */ 149 bool compress_chksum; /* compressed data chksum */ 150 unsigned char compress_ext_cnt; /* extension count */ 151 int compress_mode; /* compression mode */ 152 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 153 }; 154 155 #define F2FS_FEATURE_ENCRYPT 0x0001 156 #define F2FS_FEATURE_BLKZONED 0x0002 157 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 158 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 159 #define F2FS_FEATURE_PRJQUOTA 0x0010 160 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 161 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 162 #define F2FS_FEATURE_QUOTA_INO 0x0080 163 #define F2FS_FEATURE_INODE_CRTIME 0x0100 164 #define F2FS_FEATURE_LOST_FOUND 0x0200 165 #define F2FS_FEATURE_VERITY 0x0400 166 #define F2FS_FEATURE_SB_CHKSUM 0x0800 167 #define F2FS_FEATURE_CASEFOLD 0x1000 168 #define F2FS_FEATURE_COMPRESSION 0x2000 169 170 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 171 ((raw_super->feature & cpu_to_le32(mask)) != 0) 172 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 173 #define F2FS_SET_FEATURE(sbi, mask) \ 174 (sbi->raw_super->feature |= cpu_to_le32(mask)) 175 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 176 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 177 178 /* 179 * Default values for user and/or group using reserved blocks 180 */ 181 #define F2FS_DEF_RESUID 0 182 #define F2FS_DEF_RESGID 0 183 184 /* 185 * For checkpoint manager 186 */ 187 enum { 188 NAT_BITMAP, 189 SIT_BITMAP 190 }; 191 192 #define CP_UMOUNT 0x00000001 193 #define CP_FASTBOOT 0x00000002 194 #define CP_SYNC 0x00000004 195 #define CP_RECOVERY 0x00000008 196 #define CP_DISCARD 0x00000010 197 #define CP_TRIMMED 0x00000020 198 #define CP_PAUSE 0x00000040 199 #define CP_RESIZE 0x00000080 200 201 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 202 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 203 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 204 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 205 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 206 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 207 #define DEF_CP_INTERVAL 60 /* 60 secs */ 208 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 209 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 210 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 211 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 212 213 struct cp_control { 214 int reason; 215 __u64 trim_start; 216 __u64 trim_end; 217 __u64 trim_minlen; 218 }; 219 220 /* 221 * indicate meta/data type 222 */ 223 enum { 224 META_CP, 225 META_NAT, 226 META_SIT, 227 META_SSA, 228 META_MAX, 229 META_POR, 230 DATA_GENERIC, /* check range only */ 231 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 232 DATA_GENERIC_ENHANCE_READ, /* 233 * strong check on range and segment 234 * bitmap but no warning due to race 235 * condition of read on truncated area 236 * by extent_cache 237 */ 238 META_GENERIC, 239 }; 240 241 /* for the list of ino */ 242 enum { 243 ORPHAN_INO, /* for orphan ino list */ 244 APPEND_INO, /* for append ino list */ 245 UPDATE_INO, /* for update ino list */ 246 TRANS_DIR_INO, /* for trasactions dir ino list */ 247 FLUSH_INO, /* for multiple device flushing */ 248 MAX_INO_ENTRY, /* max. list */ 249 }; 250 251 struct ino_entry { 252 struct list_head list; /* list head */ 253 nid_t ino; /* inode number */ 254 unsigned int dirty_device; /* dirty device bitmap */ 255 }; 256 257 /* for the list of inodes to be GCed */ 258 struct inode_entry { 259 struct list_head list; /* list head */ 260 struct inode *inode; /* vfs inode pointer */ 261 }; 262 263 struct fsync_node_entry { 264 struct list_head list; /* list head */ 265 struct page *page; /* warm node page pointer */ 266 unsigned int seq_id; /* sequence id */ 267 }; 268 269 /* for the bitmap indicate blocks to be discarded */ 270 struct discard_entry { 271 struct list_head list; /* list head */ 272 block_t start_blkaddr; /* start blockaddr of current segment */ 273 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 274 }; 275 276 /* default discard granularity of inner discard thread, unit: block count */ 277 #define DEFAULT_DISCARD_GRANULARITY 16 278 279 /* max discard pend list number */ 280 #define MAX_PLIST_NUM 512 281 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 282 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 283 284 enum { 285 D_PREP, /* initial */ 286 D_PARTIAL, /* partially submitted */ 287 D_SUBMIT, /* all submitted */ 288 D_DONE, /* finished */ 289 }; 290 291 struct discard_info { 292 block_t lstart; /* logical start address */ 293 block_t len; /* length */ 294 block_t start; /* actual start address in dev */ 295 }; 296 297 struct discard_cmd { 298 struct rb_node rb_node; /* rb node located in rb-tree */ 299 union { 300 struct { 301 block_t lstart; /* logical start address */ 302 block_t len; /* length */ 303 block_t start; /* actual start address in dev */ 304 }; 305 struct discard_info di; /* discard info */ 306 307 }; 308 struct list_head list; /* command list */ 309 struct completion wait; /* compleation */ 310 struct block_device *bdev; /* bdev */ 311 unsigned short ref; /* reference count */ 312 unsigned char state; /* state */ 313 unsigned char queued; /* queued discard */ 314 int error; /* bio error */ 315 spinlock_t lock; /* for state/bio_ref updating */ 316 unsigned short bio_ref; /* bio reference count */ 317 }; 318 319 enum { 320 DPOLICY_BG, 321 DPOLICY_FORCE, 322 DPOLICY_FSTRIM, 323 DPOLICY_UMOUNT, 324 MAX_DPOLICY, 325 }; 326 327 struct discard_policy { 328 int type; /* type of discard */ 329 unsigned int min_interval; /* used for candidates exist */ 330 unsigned int mid_interval; /* used for device busy */ 331 unsigned int max_interval; /* used for candidates not exist */ 332 unsigned int max_requests; /* # of discards issued per round */ 333 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 334 bool io_aware; /* issue discard in idle time */ 335 bool sync; /* submit discard with REQ_SYNC flag */ 336 bool ordered; /* issue discard by lba order */ 337 bool timeout; /* discard timeout for put_super */ 338 unsigned int granularity; /* discard granularity */ 339 }; 340 341 struct discard_cmd_control { 342 struct task_struct *f2fs_issue_discard; /* discard thread */ 343 struct list_head entry_list; /* 4KB discard entry list */ 344 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 345 struct list_head wait_list; /* store on-flushing entries */ 346 struct list_head fstrim_list; /* in-flight discard from fstrim */ 347 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 348 unsigned int discard_wake; /* to wake up discard thread */ 349 struct mutex cmd_lock; 350 unsigned int nr_discards; /* # of discards in the list */ 351 unsigned int max_discards; /* max. discards to be issued */ 352 unsigned int discard_granularity; /* discard granularity */ 353 unsigned int undiscard_blks; /* # of undiscard blocks */ 354 unsigned int next_pos; /* next discard position */ 355 atomic_t issued_discard; /* # of issued discard */ 356 atomic_t queued_discard; /* # of queued discard */ 357 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 358 struct rb_root_cached root; /* root of discard rb-tree */ 359 bool rbtree_check; /* config for consistence check */ 360 }; 361 362 /* for the list of fsync inodes, used only during recovery */ 363 struct fsync_inode_entry { 364 struct list_head list; /* list head */ 365 struct inode *inode; /* vfs inode pointer */ 366 block_t blkaddr; /* block address locating the last fsync */ 367 block_t last_dentry; /* block address locating the last dentry */ 368 }; 369 370 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 371 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 372 373 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 374 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 375 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 376 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 377 378 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 379 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 380 381 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 382 { 383 int before = nats_in_cursum(journal); 384 385 journal->n_nats = cpu_to_le16(before + i); 386 return before; 387 } 388 389 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 390 { 391 int before = sits_in_cursum(journal); 392 393 journal->n_sits = cpu_to_le16(before + i); 394 return before; 395 } 396 397 static inline bool __has_cursum_space(struct f2fs_journal *journal, 398 int size, int type) 399 { 400 if (type == NAT_JOURNAL) 401 return size <= MAX_NAT_JENTRIES(journal); 402 return size <= MAX_SIT_JENTRIES(journal); 403 } 404 405 /* for inline stuff */ 406 #define DEF_INLINE_RESERVED_SIZE 1 407 static inline int get_extra_isize(struct inode *inode); 408 static inline int get_inline_xattr_addrs(struct inode *inode); 409 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 410 (CUR_ADDRS_PER_INODE(inode) - \ 411 get_inline_xattr_addrs(inode) - \ 412 DEF_INLINE_RESERVED_SIZE)) 413 414 /* for inline dir */ 415 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 416 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 417 BITS_PER_BYTE + 1)) 418 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 419 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 420 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 421 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 422 NR_INLINE_DENTRY(inode) + \ 423 INLINE_DENTRY_BITMAP_SIZE(inode))) 424 425 /* 426 * For INODE and NODE manager 427 */ 428 /* for directory operations */ 429 430 struct f2fs_filename { 431 /* 432 * The filename the user specified. This is NULL for some 433 * filesystem-internal operations, e.g. converting an inline directory 434 * to a non-inline one, or roll-forward recovering an encrypted dentry. 435 */ 436 const struct qstr *usr_fname; 437 438 /* 439 * The on-disk filename. For encrypted directories, this is encrypted. 440 * This may be NULL for lookups in an encrypted dir without the key. 441 */ 442 struct fscrypt_str disk_name; 443 444 /* The dirhash of this filename */ 445 f2fs_hash_t hash; 446 447 #ifdef CONFIG_FS_ENCRYPTION 448 /* 449 * For lookups in encrypted directories: either the buffer backing 450 * disk_name, or a buffer that holds the decoded no-key name. 451 */ 452 struct fscrypt_str crypto_buf; 453 #endif 454 #ifdef CONFIG_UNICODE 455 /* 456 * For casefolded directories: the casefolded name, but it's left NULL 457 * if the original name is not valid Unicode, if the directory is both 458 * casefolded and encrypted and its encryption key is unavailable, or if 459 * the filesystem is doing an internal operation where usr_fname is also 460 * NULL. In all these cases we fall back to treating the name as an 461 * opaque byte sequence. 462 */ 463 struct fscrypt_str cf_name; 464 #endif 465 }; 466 467 struct f2fs_dentry_ptr { 468 struct inode *inode; 469 void *bitmap; 470 struct f2fs_dir_entry *dentry; 471 __u8 (*filename)[F2FS_SLOT_LEN]; 472 int max; 473 int nr_bitmap; 474 }; 475 476 static inline void make_dentry_ptr_block(struct inode *inode, 477 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 478 { 479 d->inode = inode; 480 d->max = NR_DENTRY_IN_BLOCK; 481 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 482 d->bitmap = t->dentry_bitmap; 483 d->dentry = t->dentry; 484 d->filename = t->filename; 485 } 486 487 static inline void make_dentry_ptr_inline(struct inode *inode, 488 struct f2fs_dentry_ptr *d, void *t) 489 { 490 int entry_cnt = NR_INLINE_DENTRY(inode); 491 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 492 int reserved_size = INLINE_RESERVED_SIZE(inode); 493 494 d->inode = inode; 495 d->max = entry_cnt; 496 d->nr_bitmap = bitmap_size; 497 d->bitmap = t; 498 d->dentry = t + bitmap_size + reserved_size; 499 d->filename = t + bitmap_size + reserved_size + 500 SIZE_OF_DIR_ENTRY * entry_cnt; 501 } 502 503 /* 504 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 505 * as its node offset to distinguish from index node blocks. 506 * But some bits are used to mark the node block. 507 */ 508 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 509 >> OFFSET_BIT_SHIFT) 510 enum { 511 ALLOC_NODE, /* allocate a new node page if needed */ 512 LOOKUP_NODE, /* look up a node without readahead */ 513 LOOKUP_NODE_RA, /* 514 * look up a node with readahead called 515 * by get_data_block. 516 */ 517 }; 518 519 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 520 521 /* congestion wait timeout value, default: 20ms */ 522 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 523 524 /* maximum retry quota flush count */ 525 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 526 527 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 528 529 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 530 531 /* for in-memory extent cache entry */ 532 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 533 534 /* number of extent info in extent cache we try to shrink */ 535 #define EXTENT_CACHE_SHRINK_NUMBER 128 536 537 struct rb_entry { 538 struct rb_node rb_node; /* rb node located in rb-tree */ 539 union { 540 struct { 541 unsigned int ofs; /* start offset of the entry */ 542 unsigned int len; /* length of the entry */ 543 }; 544 unsigned long long key; /* 64-bits key */ 545 } __packed; 546 }; 547 548 struct extent_info { 549 unsigned int fofs; /* start offset in a file */ 550 unsigned int len; /* length of the extent */ 551 u32 blk; /* start block address of the extent */ 552 }; 553 554 struct extent_node { 555 struct rb_node rb_node; /* rb node located in rb-tree */ 556 struct extent_info ei; /* extent info */ 557 struct list_head list; /* node in global extent list of sbi */ 558 struct extent_tree *et; /* extent tree pointer */ 559 }; 560 561 struct extent_tree { 562 nid_t ino; /* inode number */ 563 struct rb_root_cached root; /* root of extent info rb-tree */ 564 struct extent_node *cached_en; /* recently accessed extent node */ 565 struct extent_info largest; /* largested extent info */ 566 struct list_head list; /* to be used by sbi->zombie_list */ 567 rwlock_t lock; /* protect extent info rb-tree */ 568 atomic_t node_cnt; /* # of extent node in rb-tree*/ 569 bool largest_updated; /* largest extent updated */ 570 }; 571 572 /* 573 * This structure is taken from ext4_map_blocks. 574 * 575 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 576 */ 577 #define F2FS_MAP_NEW (1 << BH_New) 578 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 579 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 580 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 581 F2FS_MAP_UNWRITTEN) 582 583 struct f2fs_map_blocks { 584 block_t m_pblk; 585 block_t m_lblk; 586 unsigned int m_len; 587 unsigned int m_flags; 588 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 589 pgoff_t *m_next_extent; /* point to next possible extent */ 590 int m_seg_type; 591 bool m_may_create; /* indicate it is from write path */ 592 }; 593 594 /* for flag in get_data_block */ 595 enum { 596 F2FS_GET_BLOCK_DEFAULT, 597 F2FS_GET_BLOCK_FIEMAP, 598 F2FS_GET_BLOCK_BMAP, 599 F2FS_GET_BLOCK_DIO, 600 F2FS_GET_BLOCK_PRE_DIO, 601 F2FS_GET_BLOCK_PRE_AIO, 602 F2FS_GET_BLOCK_PRECACHE, 603 }; 604 605 /* 606 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 607 */ 608 #define FADVISE_COLD_BIT 0x01 609 #define FADVISE_LOST_PINO_BIT 0x02 610 #define FADVISE_ENCRYPT_BIT 0x04 611 #define FADVISE_ENC_NAME_BIT 0x08 612 #define FADVISE_KEEP_SIZE_BIT 0x10 613 #define FADVISE_HOT_BIT 0x20 614 #define FADVISE_VERITY_BIT 0x40 615 616 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 617 618 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 619 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 620 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 621 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 622 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 623 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 624 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 625 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 626 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 627 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 628 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 629 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 630 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 631 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 632 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 633 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 634 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 635 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 636 637 #define DEF_DIR_LEVEL 0 638 639 enum { 640 GC_FAILURE_PIN, 641 GC_FAILURE_ATOMIC, 642 MAX_GC_FAILURE 643 }; 644 645 /* used for f2fs_inode_info->flags */ 646 enum { 647 FI_NEW_INODE, /* indicate newly allocated inode */ 648 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 649 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 650 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 651 FI_INC_LINK, /* need to increment i_nlink */ 652 FI_ACL_MODE, /* indicate acl mode */ 653 FI_NO_ALLOC, /* should not allocate any blocks */ 654 FI_FREE_NID, /* free allocated nide */ 655 FI_NO_EXTENT, /* not to use the extent cache */ 656 FI_INLINE_XATTR, /* used for inline xattr */ 657 FI_INLINE_DATA, /* used for inline data*/ 658 FI_INLINE_DENTRY, /* used for inline dentry */ 659 FI_APPEND_WRITE, /* inode has appended data */ 660 FI_UPDATE_WRITE, /* inode has in-place-update data */ 661 FI_NEED_IPU, /* used for ipu per file */ 662 FI_ATOMIC_FILE, /* indicate atomic file */ 663 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 664 FI_VOLATILE_FILE, /* indicate volatile file */ 665 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 666 FI_DROP_CACHE, /* drop dirty page cache */ 667 FI_DATA_EXIST, /* indicate data exists */ 668 FI_INLINE_DOTS, /* indicate inline dot dentries */ 669 FI_DO_DEFRAG, /* indicate defragment is running */ 670 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 671 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 672 FI_HOT_DATA, /* indicate file is hot */ 673 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 674 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 675 FI_PIN_FILE, /* indicate file should not be gced */ 676 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 677 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 678 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 679 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 680 FI_MMAP_FILE, /* indicate file was mmapped */ 681 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 682 FI_MAX, /* max flag, never be used */ 683 }; 684 685 struct f2fs_inode_info { 686 struct inode vfs_inode; /* serve a vfs inode */ 687 unsigned long i_flags; /* keep an inode flags for ioctl */ 688 unsigned char i_advise; /* use to give file attribute hints */ 689 unsigned char i_dir_level; /* use for dentry level for large dir */ 690 unsigned int i_current_depth; /* only for directory depth */ 691 /* for gc failure statistic */ 692 unsigned int i_gc_failures[MAX_GC_FAILURE]; 693 unsigned int i_pino; /* parent inode number */ 694 umode_t i_acl_mode; /* keep file acl mode temporarily */ 695 696 /* Use below internally in f2fs*/ 697 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 698 struct rw_semaphore i_sem; /* protect fi info */ 699 atomic_t dirty_pages; /* # of dirty pages */ 700 f2fs_hash_t chash; /* hash value of given file name */ 701 unsigned int clevel; /* maximum level of given file name */ 702 struct task_struct *task; /* lookup and create consistency */ 703 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 704 nid_t i_xattr_nid; /* node id that contains xattrs */ 705 loff_t last_disk_size; /* lastly written file size */ 706 spinlock_t i_size_lock; /* protect last_disk_size */ 707 708 #ifdef CONFIG_QUOTA 709 struct dquot *i_dquot[MAXQUOTAS]; 710 711 /* quota space reservation, managed internally by quota code */ 712 qsize_t i_reserved_quota; 713 #endif 714 struct list_head dirty_list; /* dirty list for dirs and files */ 715 struct list_head gdirty_list; /* linked in global dirty list */ 716 struct list_head inmem_ilist; /* list for inmem inodes */ 717 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 718 struct task_struct *inmem_task; /* store inmemory task */ 719 struct mutex inmem_lock; /* lock for inmemory pages */ 720 pgoff_t ra_offset; /* ongoing readahead offset */ 721 struct extent_tree *extent_tree; /* cached extent_tree entry */ 722 723 /* avoid racing between foreground op and gc */ 724 struct rw_semaphore i_gc_rwsem[2]; 725 struct rw_semaphore i_mmap_sem; 726 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 727 728 int i_extra_isize; /* size of extra space located in i_addr */ 729 kprojid_t i_projid; /* id for project quota */ 730 int i_inline_xattr_size; /* inline xattr size */ 731 struct timespec64 i_crtime; /* inode creation time */ 732 struct timespec64 i_disk_time[4];/* inode disk times */ 733 734 /* for file compress */ 735 atomic_t i_compr_blocks; /* # of compressed blocks */ 736 unsigned char i_compress_algorithm; /* algorithm type */ 737 unsigned char i_log_cluster_size; /* log of cluster size */ 738 unsigned short i_compress_flag; /* compress flag */ 739 unsigned int i_cluster_size; /* cluster size */ 740 }; 741 742 static inline void get_extent_info(struct extent_info *ext, 743 struct f2fs_extent *i_ext) 744 { 745 ext->fofs = le32_to_cpu(i_ext->fofs); 746 ext->blk = le32_to_cpu(i_ext->blk); 747 ext->len = le32_to_cpu(i_ext->len); 748 } 749 750 static inline void set_raw_extent(struct extent_info *ext, 751 struct f2fs_extent *i_ext) 752 { 753 i_ext->fofs = cpu_to_le32(ext->fofs); 754 i_ext->blk = cpu_to_le32(ext->blk); 755 i_ext->len = cpu_to_le32(ext->len); 756 } 757 758 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 759 u32 blk, unsigned int len) 760 { 761 ei->fofs = fofs; 762 ei->blk = blk; 763 ei->len = len; 764 } 765 766 static inline bool __is_discard_mergeable(struct discard_info *back, 767 struct discard_info *front, unsigned int max_len) 768 { 769 return (back->lstart + back->len == front->lstart) && 770 (back->len + front->len <= max_len); 771 } 772 773 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 774 struct discard_info *back, unsigned int max_len) 775 { 776 return __is_discard_mergeable(back, cur, max_len); 777 } 778 779 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 780 struct discard_info *front, unsigned int max_len) 781 { 782 return __is_discard_mergeable(cur, front, max_len); 783 } 784 785 static inline bool __is_extent_mergeable(struct extent_info *back, 786 struct extent_info *front) 787 { 788 return (back->fofs + back->len == front->fofs && 789 back->blk + back->len == front->blk); 790 } 791 792 static inline bool __is_back_mergeable(struct extent_info *cur, 793 struct extent_info *back) 794 { 795 return __is_extent_mergeable(back, cur); 796 } 797 798 static inline bool __is_front_mergeable(struct extent_info *cur, 799 struct extent_info *front) 800 { 801 return __is_extent_mergeable(cur, front); 802 } 803 804 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 805 static inline void __try_update_largest_extent(struct extent_tree *et, 806 struct extent_node *en) 807 { 808 if (en->ei.len > et->largest.len) { 809 et->largest = en->ei; 810 et->largest_updated = true; 811 } 812 } 813 814 /* 815 * For free nid management 816 */ 817 enum nid_state { 818 FREE_NID, /* newly added to free nid list */ 819 PREALLOC_NID, /* it is preallocated */ 820 MAX_NID_STATE, 821 }; 822 823 enum nat_state { 824 TOTAL_NAT, 825 DIRTY_NAT, 826 RECLAIMABLE_NAT, 827 MAX_NAT_STATE, 828 }; 829 830 struct f2fs_nm_info { 831 block_t nat_blkaddr; /* base disk address of NAT */ 832 nid_t max_nid; /* maximum possible node ids */ 833 nid_t available_nids; /* # of available node ids */ 834 nid_t next_scan_nid; /* the next nid to be scanned */ 835 unsigned int ram_thresh; /* control the memory footprint */ 836 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 837 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 838 839 /* NAT cache management */ 840 struct radix_tree_root nat_root;/* root of the nat entry cache */ 841 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 842 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 843 struct list_head nat_entries; /* cached nat entry list (clean) */ 844 spinlock_t nat_list_lock; /* protect clean nat entry list */ 845 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 846 unsigned int nat_blocks; /* # of nat blocks */ 847 848 /* free node ids management */ 849 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 850 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 851 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 852 spinlock_t nid_list_lock; /* protect nid lists ops */ 853 struct mutex build_lock; /* lock for build free nids */ 854 unsigned char **free_nid_bitmap; 855 unsigned char *nat_block_bitmap; 856 unsigned short *free_nid_count; /* free nid count of NAT block */ 857 858 /* for checkpoint */ 859 char *nat_bitmap; /* NAT bitmap pointer */ 860 861 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 862 unsigned char *nat_bits; /* NAT bits blocks */ 863 unsigned char *full_nat_bits; /* full NAT pages */ 864 unsigned char *empty_nat_bits; /* empty NAT pages */ 865 #ifdef CONFIG_F2FS_CHECK_FS 866 char *nat_bitmap_mir; /* NAT bitmap mirror */ 867 #endif 868 int bitmap_size; /* bitmap size */ 869 }; 870 871 /* 872 * this structure is used as one of function parameters. 873 * all the information are dedicated to a given direct node block determined 874 * by the data offset in a file. 875 */ 876 struct dnode_of_data { 877 struct inode *inode; /* vfs inode pointer */ 878 struct page *inode_page; /* its inode page, NULL is possible */ 879 struct page *node_page; /* cached direct node page */ 880 nid_t nid; /* node id of the direct node block */ 881 unsigned int ofs_in_node; /* data offset in the node page */ 882 bool inode_page_locked; /* inode page is locked or not */ 883 bool node_changed; /* is node block changed */ 884 char cur_level; /* level of hole node page */ 885 char max_level; /* level of current page located */ 886 block_t data_blkaddr; /* block address of the node block */ 887 }; 888 889 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 890 struct page *ipage, struct page *npage, nid_t nid) 891 { 892 memset(dn, 0, sizeof(*dn)); 893 dn->inode = inode; 894 dn->inode_page = ipage; 895 dn->node_page = npage; 896 dn->nid = nid; 897 } 898 899 /* 900 * For SIT manager 901 * 902 * By default, there are 6 active log areas across the whole main area. 903 * When considering hot and cold data separation to reduce cleaning overhead, 904 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 905 * respectively. 906 * In the current design, you should not change the numbers intentionally. 907 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 908 * logs individually according to the underlying devices. (default: 6) 909 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 910 * data and 8 for node logs. 911 */ 912 #define NR_CURSEG_DATA_TYPE (3) 913 #define NR_CURSEG_NODE_TYPE (3) 914 #define NR_CURSEG_INMEM_TYPE (2) 915 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 916 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 917 918 enum { 919 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 920 CURSEG_WARM_DATA, /* data blocks */ 921 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 922 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 923 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 924 CURSEG_COLD_NODE, /* indirect node blocks */ 925 NR_PERSISTENT_LOG, /* number of persistent log */ 926 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 927 /* pinned file that needs consecutive block address */ 928 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 929 NO_CHECK_TYPE, /* number of persistent & inmem log */ 930 }; 931 932 struct flush_cmd { 933 struct completion wait; 934 struct llist_node llnode; 935 nid_t ino; 936 int ret; 937 }; 938 939 struct flush_cmd_control { 940 struct task_struct *f2fs_issue_flush; /* flush thread */ 941 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 942 atomic_t issued_flush; /* # of issued flushes */ 943 atomic_t queued_flush; /* # of queued flushes */ 944 struct llist_head issue_list; /* list for command issue */ 945 struct llist_node *dispatch_list; /* list for command dispatch */ 946 }; 947 948 struct f2fs_sm_info { 949 struct sit_info *sit_info; /* whole segment information */ 950 struct free_segmap_info *free_info; /* free segment information */ 951 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 952 struct curseg_info *curseg_array; /* active segment information */ 953 954 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 955 956 block_t seg0_blkaddr; /* block address of 0'th segment */ 957 block_t main_blkaddr; /* start block address of main area */ 958 block_t ssa_blkaddr; /* start block address of SSA area */ 959 960 unsigned int segment_count; /* total # of segments */ 961 unsigned int main_segments; /* # of segments in main area */ 962 unsigned int reserved_segments; /* # of reserved segments */ 963 unsigned int ovp_segments; /* # of overprovision segments */ 964 965 /* a threshold to reclaim prefree segments */ 966 unsigned int rec_prefree_segments; 967 968 /* for batched trimming */ 969 unsigned int trim_sections; /* # of sections to trim */ 970 971 struct list_head sit_entry_set; /* sit entry set list */ 972 973 unsigned int ipu_policy; /* in-place-update policy */ 974 unsigned int min_ipu_util; /* in-place-update threshold */ 975 unsigned int min_fsync_blocks; /* threshold for fsync */ 976 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 977 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 978 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 979 980 /* for flush command control */ 981 struct flush_cmd_control *fcc_info; 982 983 /* for discard command control */ 984 struct discard_cmd_control *dcc_info; 985 }; 986 987 /* 988 * For superblock 989 */ 990 /* 991 * COUNT_TYPE for monitoring 992 * 993 * f2fs monitors the number of several block types such as on-writeback, 994 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 995 */ 996 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 997 enum count_type { 998 F2FS_DIRTY_DENTS, 999 F2FS_DIRTY_DATA, 1000 F2FS_DIRTY_QDATA, 1001 F2FS_DIRTY_NODES, 1002 F2FS_DIRTY_META, 1003 F2FS_INMEM_PAGES, 1004 F2FS_DIRTY_IMETA, 1005 F2FS_WB_CP_DATA, 1006 F2FS_WB_DATA, 1007 F2FS_RD_DATA, 1008 F2FS_RD_NODE, 1009 F2FS_RD_META, 1010 F2FS_DIO_WRITE, 1011 F2FS_DIO_READ, 1012 NR_COUNT_TYPE, 1013 }; 1014 1015 /* 1016 * The below are the page types of bios used in submit_bio(). 1017 * The available types are: 1018 * DATA User data pages. It operates as async mode. 1019 * NODE Node pages. It operates as async mode. 1020 * META FS metadata pages such as SIT, NAT, CP. 1021 * NR_PAGE_TYPE The number of page types. 1022 * META_FLUSH Make sure the previous pages are written 1023 * with waiting the bio's completion 1024 * ... Only can be used with META. 1025 */ 1026 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1027 enum page_type { 1028 DATA, 1029 NODE, 1030 META, 1031 NR_PAGE_TYPE, 1032 META_FLUSH, 1033 INMEM, /* the below types are used by tracepoints only. */ 1034 INMEM_DROP, 1035 INMEM_INVALIDATE, 1036 INMEM_REVOKE, 1037 IPU, 1038 OPU, 1039 }; 1040 1041 enum temp_type { 1042 HOT = 0, /* must be zero for meta bio */ 1043 WARM, 1044 COLD, 1045 NR_TEMP_TYPE, 1046 }; 1047 1048 enum need_lock_type { 1049 LOCK_REQ = 0, 1050 LOCK_DONE, 1051 LOCK_RETRY, 1052 }; 1053 1054 enum cp_reason_type { 1055 CP_NO_NEEDED, 1056 CP_NON_REGULAR, 1057 CP_COMPRESSED, 1058 CP_HARDLINK, 1059 CP_SB_NEED_CP, 1060 CP_WRONG_PINO, 1061 CP_NO_SPC_ROLL, 1062 CP_NODE_NEED_CP, 1063 CP_FASTBOOT_MODE, 1064 CP_SPEC_LOG_NUM, 1065 CP_RECOVER_DIR, 1066 }; 1067 1068 enum iostat_type { 1069 /* WRITE IO */ 1070 APP_DIRECT_IO, /* app direct write IOs */ 1071 APP_BUFFERED_IO, /* app buffered write IOs */ 1072 APP_WRITE_IO, /* app write IOs */ 1073 APP_MAPPED_IO, /* app mapped IOs */ 1074 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1075 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1076 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1077 FS_GC_DATA_IO, /* data IOs from forground gc */ 1078 FS_GC_NODE_IO, /* node IOs from forground gc */ 1079 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1080 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1081 FS_CP_META_IO, /* meta IOs from checkpoint */ 1082 1083 /* READ IO */ 1084 APP_DIRECT_READ_IO, /* app direct read IOs */ 1085 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1086 APP_READ_IO, /* app read IOs */ 1087 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1088 FS_DATA_READ_IO, /* data read IOs */ 1089 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1090 FS_CDATA_READ_IO, /* compressed data read IOs */ 1091 FS_NODE_READ_IO, /* node read IOs */ 1092 FS_META_READ_IO, /* meta read IOs */ 1093 1094 /* other */ 1095 FS_DISCARD, /* discard */ 1096 NR_IO_TYPE, 1097 }; 1098 1099 struct f2fs_io_info { 1100 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1101 nid_t ino; /* inode number */ 1102 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1103 enum temp_type temp; /* contains HOT/WARM/COLD */ 1104 int op; /* contains REQ_OP_ */ 1105 int op_flags; /* req_flag_bits */ 1106 block_t new_blkaddr; /* new block address to be written */ 1107 block_t old_blkaddr; /* old block address before Cow */ 1108 struct page *page; /* page to be written */ 1109 struct page *encrypted_page; /* encrypted page */ 1110 struct page *compressed_page; /* compressed page */ 1111 struct list_head list; /* serialize IOs */ 1112 bool submitted; /* indicate IO submission */ 1113 int need_lock; /* indicate we need to lock cp_rwsem */ 1114 bool in_list; /* indicate fio is in io_list */ 1115 bool is_por; /* indicate IO is from recovery or not */ 1116 bool retry; /* need to reallocate block address */ 1117 int compr_blocks; /* # of compressed block addresses */ 1118 bool encrypted; /* indicate file is encrypted */ 1119 enum iostat_type io_type; /* io type */ 1120 struct writeback_control *io_wbc; /* writeback control */ 1121 struct bio **bio; /* bio for ipu */ 1122 sector_t *last_block; /* last block number in bio */ 1123 unsigned char version; /* version of the node */ 1124 }; 1125 1126 struct bio_entry { 1127 struct bio *bio; 1128 struct list_head list; 1129 }; 1130 1131 #define is_read_io(rw) ((rw) == READ) 1132 struct f2fs_bio_info { 1133 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1134 struct bio *bio; /* bios to merge */ 1135 sector_t last_block_in_bio; /* last block number */ 1136 struct f2fs_io_info fio; /* store buffered io info. */ 1137 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1138 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1139 struct list_head io_list; /* track fios */ 1140 struct list_head bio_list; /* bio entry list head */ 1141 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1142 }; 1143 1144 #define FDEV(i) (sbi->devs[i]) 1145 #define RDEV(i) (raw_super->devs[i]) 1146 struct f2fs_dev_info { 1147 struct block_device *bdev; 1148 char path[MAX_PATH_LEN]; 1149 unsigned int total_segments; 1150 block_t start_blk; 1151 block_t end_blk; 1152 #ifdef CONFIG_BLK_DEV_ZONED 1153 unsigned int nr_blkz; /* Total number of zones */ 1154 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1155 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1156 #endif 1157 }; 1158 1159 enum inode_type { 1160 DIR_INODE, /* for dirty dir inode */ 1161 FILE_INODE, /* for dirty regular/symlink inode */ 1162 DIRTY_META, /* for all dirtied inode metadata */ 1163 ATOMIC_FILE, /* for all atomic files */ 1164 NR_INODE_TYPE, 1165 }; 1166 1167 /* for inner inode cache management */ 1168 struct inode_management { 1169 struct radix_tree_root ino_root; /* ino entry array */ 1170 spinlock_t ino_lock; /* for ino entry lock */ 1171 struct list_head ino_list; /* inode list head */ 1172 unsigned long ino_num; /* number of entries */ 1173 }; 1174 1175 /* for GC_AT */ 1176 struct atgc_management { 1177 bool atgc_enabled; /* ATGC is enabled or not */ 1178 struct rb_root_cached root; /* root of victim rb-tree */ 1179 struct list_head victim_list; /* linked with all victim entries */ 1180 unsigned int victim_count; /* victim count in rb-tree */ 1181 unsigned int candidate_ratio; /* candidate ratio */ 1182 unsigned int max_candidate_count; /* max candidate count */ 1183 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1184 unsigned long long age_threshold; /* age threshold */ 1185 }; 1186 1187 /* For s_flag in struct f2fs_sb_info */ 1188 enum { 1189 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1190 SBI_IS_CLOSE, /* specify unmounting */ 1191 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1192 SBI_POR_DOING, /* recovery is doing or not */ 1193 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1194 SBI_NEED_CP, /* need to checkpoint */ 1195 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1196 SBI_IS_RECOVERED, /* recovered orphan/data */ 1197 SBI_CP_DISABLED, /* CP was disabled last mount */ 1198 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1199 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1200 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1201 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1202 SBI_IS_RESIZEFS, /* resizefs is in process */ 1203 }; 1204 1205 enum { 1206 CP_TIME, 1207 REQ_TIME, 1208 DISCARD_TIME, 1209 GC_TIME, 1210 DISABLE_TIME, 1211 UMOUNT_DISCARD_TIMEOUT, 1212 MAX_TIME, 1213 }; 1214 1215 enum { 1216 GC_NORMAL, 1217 GC_IDLE_CB, 1218 GC_IDLE_GREEDY, 1219 GC_IDLE_AT, 1220 GC_URGENT_HIGH, 1221 GC_URGENT_LOW, 1222 }; 1223 1224 enum { 1225 BGGC_MODE_ON, /* background gc is on */ 1226 BGGC_MODE_OFF, /* background gc is off */ 1227 BGGC_MODE_SYNC, /* 1228 * background gc is on, migrating blocks 1229 * like foreground gc 1230 */ 1231 }; 1232 1233 enum { 1234 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1235 FS_MODE_LFS, /* use lfs allocation only */ 1236 }; 1237 1238 enum { 1239 WHINT_MODE_OFF, /* not pass down write hints */ 1240 WHINT_MODE_USER, /* try to pass down hints given by users */ 1241 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1242 }; 1243 1244 enum { 1245 ALLOC_MODE_DEFAULT, /* stay default */ 1246 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1247 }; 1248 1249 enum fsync_mode { 1250 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1251 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1252 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1253 }; 1254 1255 enum { 1256 COMPR_MODE_FS, /* 1257 * automatically compress compression 1258 * enabled files 1259 */ 1260 COMPR_MODE_USER, /* 1261 * automatical compression is disabled. 1262 * user can control the file compression 1263 * using ioctls 1264 */ 1265 }; 1266 1267 /* 1268 * this value is set in page as a private data which indicate that 1269 * the page is atomically written, and it is in inmem_pages list. 1270 */ 1271 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1272 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1273 1274 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1275 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1276 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1277 (page_private(page) == DUMMY_WRITTEN_PAGE) 1278 1279 #ifdef CONFIG_F2FS_IO_TRACE 1280 #define IS_IO_TRACED_PAGE(page) \ 1281 (page_private(page) > 0 && \ 1282 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1283 #else 1284 #define IS_IO_TRACED_PAGE(page) (0) 1285 #endif 1286 1287 /* For compression */ 1288 enum compress_algorithm_type { 1289 COMPRESS_LZO, 1290 COMPRESS_LZ4, 1291 COMPRESS_ZSTD, 1292 COMPRESS_LZORLE, 1293 COMPRESS_MAX, 1294 }; 1295 1296 enum compress_flag { 1297 COMPRESS_CHKSUM, 1298 COMPRESS_MAX_FLAG, 1299 }; 1300 1301 #define COMPRESS_DATA_RESERVED_SIZE 4 1302 struct compress_data { 1303 __le32 clen; /* compressed data size */ 1304 __le32 chksum; /* compressed data chksum */ 1305 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1306 u8 cdata[]; /* compressed data */ 1307 }; 1308 1309 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1310 1311 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1312 1313 /* compress context */ 1314 struct compress_ctx { 1315 struct inode *inode; /* inode the context belong to */ 1316 pgoff_t cluster_idx; /* cluster index number */ 1317 unsigned int cluster_size; /* page count in cluster */ 1318 unsigned int log_cluster_size; /* log of cluster size */ 1319 struct page **rpages; /* pages store raw data in cluster */ 1320 unsigned int nr_rpages; /* total page number in rpages */ 1321 struct page **cpages; /* pages store compressed data in cluster */ 1322 unsigned int nr_cpages; /* total page number in cpages */ 1323 void *rbuf; /* virtual mapped address on rpages */ 1324 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1325 size_t rlen; /* valid data length in rbuf */ 1326 size_t clen; /* valid data length in cbuf */ 1327 void *private; /* payload buffer for specified compression algorithm */ 1328 void *private2; /* extra payload buffer */ 1329 }; 1330 1331 /* compress context for write IO path */ 1332 struct compress_io_ctx { 1333 u32 magic; /* magic number to indicate page is compressed */ 1334 struct inode *inode; /* inode the context belong to */ 1335 struct page **rpages; /* pages store raw data in cluster */ 1336 unsigned int nr_rpages; /* total page number in rpages */ 1337 atomic_t pending_pages; /* in-flight compressed page count */ 1338 }; 1339 1340 /* decompress io context for read IO path */ 1341 struct decompress_io_ctx { 1342 u32 magic; /* magic number to indicate page is compressed */ 1343 struct inode *inode; /* inode the context belong to */ 1344 pgoff_t cluster_idx; /* cluster index number */ 1345 unsigned int cluster_size; /* page count in cluster */ 1346 unsigned int log_cluster_size; /* log of cluster size */ 1347 struct page **rpages; /* pages store raw data in cluster */ 1348 unsigned int nr_rpages; /* total page number in rpages */ 1349 struct page **cpages; /* pages store compressed data in cluster */ 1350 unsigned int nr_cpages; /* total page number in cpages */ 1351 struct page **tpages; /* temp pages to pad holes in cluster */ 1352 void *rbuf; /* virtual mapped address on rpages */ 1353 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1354 size_t rlen; /* valid data length in rbuf */ 1355 size_t clen; /* valid data length in cbuf */ 1356 atomic_t pending_pages; /* in-flight compressed page count */ 1357 bool failed; /* indicate IO error during decompression */ 1358 void *private; /* payload buffer for specified decompression algorithm */ 1359 void *private2; /* extra payload buffer */ 1360 }; 1361 1362 #define NULL_CLUSTER ((unsigned int)(~0)) 1363 #define MIN_COMPRESS_LOG_SIZE 2 1364 #define MAX_COMPRESS_LOG_SIZE 8 1365 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1366 1367 struct f2fs_sb_info { 1368 struct super_block *sb; /* pointer to VFS super block */ 1369 struct proc_dir_entry *s_proc; /* proc entry */ 1370 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1371 struct rw_semaphore sb_lock; /* lock for raw super block */ 1372 int valid_super_block; /* valid super block no */ 1373 unsigned long s_flag; /* flags for sbi */ 1374 struct mutex writepages; /* mutex for writepages() */ 1375 1376 #ifdef CONFIG_BLK_DEV_ZONED 1377 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1378 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1379 #endif 1380 1381 /* for node-related operations */ 1382 struct f2fs_nm_info *nm_info; /* node manager */ 1383 struct inode *node_inode; /* cache node blocks */ 1384 1385 /* for segment-related operations */ 1386 struct f2fs_sm_info *sm_info; /* segment manager */ 1387 1388 /* for bio operations */ 1389 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1390 /* keep migration IO order for LFS mode */ 1391 struct rw_semaphore io_order_lock; 1392 mempool_t *write_io_dummy; /* Dummy pages */ 1393 1394 /* for checkpoint */ 1395 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1396 int cur_cp_pack; /* remain current cp pack */ 1397 spinlock_t cp_lock; /* for flag in ckpt */ 1398 struct inode *meta_inode; /* cache meta blocks */ 1399 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1400 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1401 struct rw_semaphore node_write; /* locking node writes */ 1402 struct rw_semaphore node_change; /* locking node change */ 1403 wait_queue_head_t cp_wait; 1404 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1405 long interval_time[MAX_TIME]; /* to store thresholds */ 1406 1407 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1408 1409 spinlock_t fsync_node_lock; /* for node entry lock */ 1410 struct list_head fsync_node_list; /* node list head */ 1411 unsigned int fsync_seg_id; /* sequence id */ 1412 unsigned int fsync_node_num; /* number of node entries */ 1413 1414 /* for orphan inode, use 0'th array */ 1415 unsigned int max_orphans; /* max orphan inodes */ 1416 1417 /* for inode management */ 1418 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1419 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1420 struct mutex flush_lock; /* for flush exclusion */ 1421 1422 /* for extent tree cache */ 1423 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1424 struct mutex extent_tree_lock; /* locking extent radix tree */ 1425 struct list_head extent_list; /* lru list for shrinker */ 1426 spinlock_t extent_lock; /* locking extent lru list */ 1427 atomic_t total_ext_tree; /* extent tree count */ 1428 struct list_head zombie_list; /* extent zombie tree list */ 1429 atomic_t total_zombie_tree; /* extent zombie tree count */ 1430 atomic_t total_ext_node; /* extent info count */ 1431 1432 /* basic filesystem units */ 1433 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1434 unsigned int log_blocksize; /* log2 block size */ 1435 unsigned int blocksize; /* block size */ 1436 unsigned int root_ino_num; /* root inode number*/ 1437 unsigned int node_ino_num; /* node inode number*/ 1438 unsigned int meta_ino_num; /* meta inode number*/ 1439 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1440 unsigned int blocks_per_seg; /* blocks per segment */ 1441 unsigned int segs_per_sec; /* segments per section */ 1442 unsigned int secs_per_zone; /* sections per zone */ 1443 unsigned int total_sections; /* total section count */ 1444 unsigned int total_node_count; /* total node block count */ 1445 unsigned int total_valid_node_count; /* valid node block count */ 1446 loff_t max_file_blocks; /* max block index of file */ 1447 int dir_level; /* directory level */ 1448 int readdir_ra; /* readahead inode in readdir */ 1449 1450 block_t user_block_count; /* # of user blocks */ 1451 block_t total_valid_block_count; /* # of valid blocks */ 1452 block_t discard_blks; /* discard command candidats */ 1453 block_t last_valid_block_count; /* for recovery */ 1454 block_t reserved_blocks; /* configurable reserved blocks */ 1455 block_t current_reserved_blocks; /* current reserved blocks */ 1456 1457 /* Additional tracking for no checkpoint mode */ 1458 block_t unusable_block_count; /* # of blocks saved by last cp */ 1459 1460 unsigned int nquota_files; /* # of quota sysfile */ 1461 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1462 1463 /* # of pages, see count_type */ 1464 atomic_t nr_pages[NR_COUNT_TYPE]; 1465 /* # of allocated blocks */ 1466 struct percpu_counter alloc_valid_block_count; 1467 1468 /* writeback control */ 1469 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1470 1471 /* valid inode count */ 1472 struct percpu_counter total_valid_inode_count; 1473 1474 struct f2fs_mount_info mount_opt; /* mount options */ 1475 1476 /* for cleaning operations */ 1477 struct rw_semaphore gc_lock; /* 1478 * semaphore for GC, avoid 1479 * race between GC and GC or CP 1480 */ 1481 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1482 struct atgc_management am; /* atgc management */ 1483 unsigned int cur_victim_sec; /* current victim section num */ 1484 unsigned int gc_mode; /* current GC state */ 1485 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1486 1487 /* for skip statistic */ 1488 unsigned int atomic_files; /* # of opened atomic file */ 1489 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1490 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1491 1492 /* threshold for gc trials on pinned files */ 1493 u64 gc_pin_file_threshold; 1494 struct rw_semaphore pin_sem; 1495 1496 /* maximum # of trials to find a victim segment for SSR and GC */ 1497 unsigned int max_victim_search; 1498 /* migration granularity of garbage collection, unit: segment */ 1499 unsigned int migration_granularity; 1500 1501 /* 1502 * for stat information. 1503 * one is for the LFS mode, and the other is for the SSR mode. 1504 */ 1505 #ifdef CONFIG_F2FS_STAT_FS 1506 struct f2fs_stat_info *stat_info; /* FS status information */ 1507 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1508 unsigned int segment_count[2]; /* # of allocated segments */ 1509 unsigned int block_count[2]; /* # of allocated blocks */ 1510 atomic_t inplace_count; /* # of inplace update */ 1511 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1512 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1513 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1514 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1515 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1516 atomic_t inline_inode; /* # of inline_data inodes */ 1517 atomic_t inline_dir; /* # of inline_dentry inodes */ 1518 atomic_t compr_inode; /* # of compressed inodes */ 1519 atomic64_t compr_blocks; /* # of compressed blocks */ 1520 atomic_t vw_cnt; /* # of volatile writes */ 1521 atomic_t max_aw_cnt; /* max # of atomic writes */ 1522 atomic_t max_vw_cnt; /* max # of volatile writes */ 1523 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1524 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1525 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1526 #endif 1527 spinlock_t stat_lock; /* lock for stat operations */ 1528 1529 /* For app/fs IO statistics */ 1530 spinlock_t iostat_lock; 1531 unsigned long long rw_iostat[NR_IO_TYPE]; 1532 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1533 bool iostat_enable; 1534 unsigned long iostat_next_period; 1535 unsigned int iostat_period_ms; 1536 1537 /* to attach REQ_META|REQ_FUA flags */ 1538 unsigned int data_io_flag; 1539 unsigned int node_io_flag; 1540 1541 /* For sysfs suppport */ 1542 struct kobject s_kobj; 1543 struct completion s_kobj_unregister; 1544 1545 /* For shrinker support */ 1546 struct list_head s_list; 1547 int s_ndevs; /* number of devices */ 1548 struct f2fs_dev_info *devs; /* for device list */ 1549 unsigned int dirty_device; /* for checkpoint data flush */ 1550 spinlock_t dev_lock; /* protect dirty_device */ 1551 struct mutex umount_mutex; 1552 unsigned int shrinker_run_no; 1553 1554 /* For write statistics */ 1555 u64 sectors_written_start; 1556 u64 kbytes_written; 1557 1558 /* Reference to checksum algorithm driver via cryptoapi */ 1559 struct crypto_shash *s_chksum_driver; 1560 1561 /* Precomputed FS UUID checksum for seeding other checksums */ 1562 __u32 s_chksum_seed; 1563 1564 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1565 1566 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1567 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1568 1569 #ifdef CONFIG_F2FS_FS_COMPRESSION 1570 struct kmem_cache *page_array_slab; /* page array entry */ 1571 unsigned int page_array_slab_size; /* default page array slab size */ 1572 #endif 1573 }; 1574 1575 struct f2fs_private_dio { 1576 struct inode *inode; 1577 void *orig_private; 1578 bio_end_io_t *orig_end_io; 1579 bool write; 1580 }; 1581 1582 #ifdef CONFIG_F2FS_FAULT_INJECTION 1583 #define f2fs_show_injection_info(sbi, type) \ 1584 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1585 KERN_INFO, sbi->sb->s_id, \ 1586 f2fs_fault_name[type], \ 1587 __func__, __builtin_return_address(0)) 1588 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1589 { 1590 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1591 1592 if (!ffi->inject_rate) 1593 return false; 1594 1595 if (!IS_FAULT_SET(ffi, type)) 1596 return false; 1597 1598 atomic_inc(&ffi->inject_ops); 1599 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1600 atomic_set(&ffi->inject_ops, 0); 1601 return true; 1602 } 1603 return false; 1604 } 1605 #else 1606 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1607 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1608 { 1609 return false; 1610 } 1611 #endif 1612 1613 /* 1614 * Test if the mounted volume is a multi-device volume. 1615 * - For a single regular disk volume, sbi->s_ndevs is 0. 1616 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1617 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1618 */ 1619 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1620 { 1621 return sbi->s_ndevs > 1; 1622 } 1623 1624 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1625 { 1626 unsigned long now = jiffies; 1627 1628 sbi->last_time[type] = now; 1629 1630 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1631 if (type == REQ_TIME) { 1632 sbi->last_time[DISCARD_TIME] = now; 1633 sbi->last_time[GC_TIME] = now; 1634 } 1635 } 1636 1637 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1638 { 1639 unsigned long interval = sbi->interval_time[type] * HZ; 1640 1641 return time_after(jiffies, sbi->last_time[type] + interval); 1642 } 1643 1644 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1645 int type) 1646 { 1647 unsigned long interval = sbi->interval_time[type] * HZ; 1648 unsigned int wait_ms = 0; 1649 long delta; 1650 1651 delta = (sbi->last_time[type] + interval) - jiffies; 1652 if (delta > 0) 1653 wait_ms = jiffies_to_msecs(delta); 1654 1655 return wait_ms; 1656 } 1657 1658 /* 1659 * Inline functions 1660 */ 1661 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1662 const void *address, unsigned int length) 1663 { 1664 struct { 1665 struct shash_desc shash; 1666 char ctx[4]; 1667 } desc; 1668 int err; 1669 1670 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1671 1672 desc.shash.tfm = sbi->s_chksum_driver; 1673 *(u32 *)desc.ctx = crc; 1674 1675 err = crypto_shash_update(&desc.shash, address, length); 1676 BUG_ON(err); 1677 1678 return *(u32 *)desc.ctx; 1679 } 1680 1681 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1682 unsigned int length) 1683 { 1684 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1685 } 1686 1687 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1688 void *buf, size_t buf_size) 1689 { 1690 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1691 } 1692 1693 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1694 const void *address, unsigned int length) 1695 { 1696 return __f2fs_crc32(sbi, crc, address, length); 1697 } 1698 1699 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1700 { 1701 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1702 } 1703 1704 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1705 { 1706 return sb->s_fs_info; 1707 } 1708 1709 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1710 { 1711 return F2FS_SB(inode->i_sb); 1712 } 1713 1714 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1715 { 1716 return F2FS_I_SB(mapping->host); 1717 } 1718 1719 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1720 { 1721 return F2FS_M_SB(page_file_mapping(page)); 1722 } 1723 1724 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1725 { 1726 return (struct f2fs_super_block *)(sbi->raw_super); 1727 } 1728 1729 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1730 { 1731 return (struct f2fs_checkpoint *)(sbi->ckpt); 1732 } 1733 1734 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1735 { 1736 return (struct f2fs_node *)page_address(page); 1737 } 1738 1739 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1740 { 1741 return &((struct f2fs_node *)page_address(page))->i; 1742 } 1743 1744 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1745 { 1746 return (struct f2fs_nm_info *)(sbi->nm_info); 1747 } 1748 1749 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1750 { 1751 return (struct f2fs_sm_info *)(sbi->sm_info); 1752 } 1753 1754 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1755 { 1756 return (struct sit_info *)(SM_I(sbi)->sit_info); 1757 } 1758 1759 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1760 { 1761 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1762 } 1763 1764 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1765 { 1766 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1767 } 1768 1769 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1770 { 1771 return sbi->meta_inode->i_mapping; 1772 } 1773 1774 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1775 { 1776 return sbi->node_inode->i_mapping; 1777 } 1778 1779 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1780 { 1781 return test_bit(type, &sbi->s_flag); 1782 } 1783 1784 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1785 { 1786 set_bit(type, &sbi->s_flag); 1787 } 1788 1789 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1790 { 1791 clear_bit(type, &sbi->s_flag); 1792 } 1793 1794 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1795 { 1796 return le64_to_cpu(cp->checkpoint_ver); 1797 } 1798 1799 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1800 { 1801 if (type < F2FS_MAX_QUOTAS) 1802 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1803 return 0; 1804 } 1805 1806 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1807 { 1808 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1809 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1810 } 1811 1812 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1813 { 1814 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1815 1816 return ckpt_flags & f; 1817 } 1818 1819 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1820 { 1821 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1822 } 1823 1824 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1825 { 1826 unsigned int ckpt_flags; 1827 1828 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1829 ckpt_flags |= f; 1830 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1831 } 1832 1833 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1834 { 1835 unsigned long flags; 1836 1837 spin_lock_irqsave(&sbi->cp_lock, flags); 1838 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1839 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1840 } 1841 1842 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1843 { 1844 unsigned int ckpt_flags; 1845 1846 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1847 ckpt_flags &= (~f); 1848 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1849 } 1850 1851 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1852 { 1853 unsigned long flags; 1854 1855 spin_lock_irqsave(&sbi->cp_lock, flags); 1856 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1857 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1858 } 1859 1860 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1861 { 1862 unsigned long flags; 1863 unsigned char *nat_bits; 1864 1865 /* 1866 * In order to re-enable nat_bits we need to call fsck.f2fs by 1867 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1868 * so let's rely on regular fsck or unclean shutdown. 1869 */ 1870 1871 if (lock) 1872 spin_lock_irqsave(&sbi->cp_lock, flags); 1873 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1874 nat_bits = NM_I(sbi)->nat_bits; 1875 NM_I(sbi)->nat_bits = NULL; 1876 if (lock) 1877 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1878 1879 kvfree(nat_bits); 1880 } 1881 1882 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1883 struct cp_control *cpc) 1884 { 1885 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1886 1887 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1888 } 1889 1890 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1891 { 1892 down_read(&sbi->cp_rwsem); 1893 } 1894 1895 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1896 { 1897 return down_read_trylock(&sbi->cp_rwsem); 1898 } 1899 1900 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1901 { 1902 up_read(&sbi->cp_rwsem); 1903 } 1904 1905 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1906 { 1907 down_write(&sbi->cp_rwsem); 1908 } 1909 1910 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1911 { 1912 up_write(&sbi->cp_rwsem); 1913 } 1914 1915 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1916 { 1917 int reason = CP_SYNC; 1918 1919 if (test_opt(sbi, FASTBOOT)) 1920 reason = CP_FASTBOOT; 1921 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1922 reason = CP_UMOUNT; 1923 return reason; 1924 } 1925 1926 static inline bool __remain_node_summaries(int reason) 1927 { 1928 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1929 } 1930 1931 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1932 { 1933 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1934 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1935 } 1936 1937 /* 1938 * Check whether the inode has blocks or not 1939 */ 1940 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1941 { 1942 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1943 1944 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1945 } 1946 1947 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1948 { 1949 return ofs == XATTR_NODE_OFFSET; 1950 } 1951 1952 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1953 struct inode *inode, bool cap) 1954 { 1955 if (!inode) 1956 return true; 1957 if (!test_opt(sbi, RESERVE_ROOT)) 1958 return false; 1959 if (IS_NOQUOTA(inode)) 1960 return true; 1961 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1962 return true; 1963 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1964 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1965 return true; 1966 if (cap && capable(CAP_SYS_RESOURCE)) 1967 return true; 1968 return false; 1969 } 1970 1971 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1972 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1973 struct inode *inode, blkcnt_t *count) 1974 { 1975 blkcnt_t diff = 0, release = 0; 1976 block_t avail_user_block_count; 1977 int ret; 1978 1979 ret = dquot_reserve_block(inode, *count); 1980 if (ret) 1981 return ret; 1982 1983 if (time_to_inject(sbi, FAULT_BLOCK)) { 1984 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1985 release = *count; 1986 goto release_quota; 1987 } 1988 1989 /* 1990 * let's increase this in prior to actual block count change in order 1991 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1992 */ 1993 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1994 1995 spin_lock(&sbi->stat_lock); 1996 sbi->total_valid_block_count += (block_t)(*count); 1997 avail_user_block_count = sbi->user_block_count - 1998 sbi->current_reserved_blocks; 1999 2000 if (!__allow_reserved_blocks(sbi, inode, true)) 2001 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2002 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2003 if (avail_user_block_count > sbi->unusable_block_count) 2004 avail_user_block_count -= sbi->unusable_block_count; 2005 else 2006 avail_user_block_count = 0; 2007 } 2008 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2009 diff = sbi->total_valid_block_count - avail_user_block_count; 2010 if (diff > *count) 2011 diff = *count; 2012 *count -= diff; 2013 release = diff; 2014 sbi->total_valid_block_count -= diff; 2015 if (!*count) { 2016 spin_unlock(&sbi->stat_lock); 2017 goto enospc; 2018 } 2019 } 2020 spin_unlock(&sbi->stat_lock); 2021 2022 if (unlikely(release)) { 2023 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2024 dquot_release_reservation_block(inode, release); 2025 } 2026 f2fs_i_blocks_write(inode, *count, true, true); 2027 return 0; 2028 2029 enospc: 2030 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2031 release_quota: 2032 dquot_release_reservation_block(inode, release); 2033 return -ENOSPC; 2034 } 2035 2036 __printf(2, 3) 2037 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2038 2039 #define f2fs_err(sbi, fmt, ...) \ 2040 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2041 #define f2fs_warn(sbi, fmt, ...) \ 2042 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2043 #define f2fs_notice(sbi, fmt, ...) \ 2044 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2045 #define f2fs_info(sbi, fmt, ...) \ 2046 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2047 #define f2fs_debug(sbi, fmt, ...) \ 2048 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2049 2050 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2051 struct inode *inode, 2052 block_t count) 2053 { 2054 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2055 2056 spin_lock(&sbi->stat_lock); 2057 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2058 sbi->total_valid_block_count -= (block_t)count; 2059 if (sbi->reserved_blocks && 2060 sbi->current_reserved_blocks < sbi->reserved_blocks) 2061 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2062 sbi->current_reserved_blocks + count); 2063 spin_unlock(&sbi->stat_lock); 2064 if (unlikely(inode->i_blocks < sectors)) { 2065 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2066 inode->i_ino, 2067 (unsigned long long)inode->i_blocks, 2068 (unsigned long long)sectors); 2069 set_sbi_flag(sbi, SBI_NEED_FSCK); 2070 return; 2071 } 2072 f2fs_i_blocks_write(inode, count, false, true); 2073 } 2074 2075 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2076 { 2077 atomic_inc(&sbi->nr_pages[count_type]); 2078 2079 if (count_type == F2FS_DIRTY_DENTS || 2080 count_type == F2FS_DIRTY_NODES || 2081 count_type == F2FS_DIRTY_META || 2082 count_type == F2FS_DIRTY_QDATA || 2083 count_type == F2FS_DIRTY_IMETA) 2084 set_sbi_flag(sbi, SBI_IS_DIRTY); 2085 } 2086 2087 static inline void inode_inc_dirty_pages(struct inode *inode) 2088 { 2089 atomic_inc(&F2FS_I(inode)->dirty_pages); 2090 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2091 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2092 if (IS_NOQUOTA(inode)) 2093 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2094 } 2095 2096 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2097 { 2098 atomic_dec(&sbi->nr_pages[count_type]); 2099 } 2100 2101 static inline void inode_dec_dirty_pages(struct inode *inode) 2102 { 2103 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2104 !S_ISLNK(inode->i_mode)) 2105 return; 2106 2107 atomic_dec(&F2FS_I(inode)->dirty_pages); 2108 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2109 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2110 if (IS_NOQUOTA(inode)) 2111 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2112 } 2113 2114 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2115 { 2116 return atomic_read(&sbi->nr_pages[count_type]); 2117 } 2118 2119 static inline int get_dirty_pages(struct inode *inode) 2120 { 2121 return atomic_read(&F2FS_I(inode)->dirty_pages); 2122 } 2123 2124 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2125 { 2126 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2127 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2128 sbi->log_blocks_per_seg; 2129 2130 return segs / sbi->segs_per_sec; 2131 } 2132 2133 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2134 { 2135 return sbi->total_valid_block_count; 2136 } 2137 2138 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2139 { 2140 return sbi->discard_blks; 2141 } 2142 2143 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2144 { 2145 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2146 2147 /* return NAT or SIT bitmap */ 2148 if (flag == NAT_BITMAP) 2149 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2150 else if (flag == SIT_BITMAP) 2151 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2152 2153 return 0; 2154 } 2155 2156 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2157 { 2158 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2159 } 2160 2161 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2162 { 2163 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2164 int offset; 2165 2166 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2167 offset = (flag == SIT_BITMAP) ? 2168 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2169 /* 2170 * if large_nat_bitmap feature is enabled, leave checksum 2171 * protection for all nat/sit bitmaps. 2172 */ 2173 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2174 } 2175 2176 if (__cp_payload(sbi) > 0) { 2177 if (flag == NAT_BITMAP) 2178 return &ckpt->sit_nat_version_bitmap; 2179 else 2180 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2181 } else { 2182 offset = (flag == NAT_BITMAP) ? 2183 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2184 return &ckpt->sit_nat_version_bitmap + offset; 2185 } 2186 } 2187 2188 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2189 { 2190 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2191 2192 if (sbi->cur_cp_pack == 2) 2193 start_addr += sbi->blocks_per_seg; 2194 return start_addr; 2195 } 2196 2197 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2198 { 2199 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2200 2201 if (sbi->cur_cp_pack == 1) 2202 start_addr += sbi->blocks_per_seg; 2203 return start_addr; 2204 } 2205 2206 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2207 { 2208 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2209 } 2210 2211 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2212 { 2213 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2214 } 2215 2216 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2217 struct inode *inode, bool is_inode) 2218 { 2219 block_t valid_block_count; 2220 unsigned int valid_node_count, user_block_count; 2221 int err; 2222 2223 if (is_inode) { 2224 if (inode) { 2225 err = dquot_alloc_inode(inode); 2226 if (err) 2227 return err; 2228 } 2229 } else { 2230 err = dquot_reserve_block(inode, 1); 2231 if (err) 2232 return err; 2233 } 2234 2235 if (time_to_inject(sbi, FAULT_BLOCK)) { 2236 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2237 goto enospc; 2238 } 2239 2240 spin_lock(&sbi->stat_lock); 2241 2242 valid_block_count = sbi->total_valid_block_count + 2243 sbi->current_reserved_blocks + 1; 2244 2245 if (!__allow_reserved_blocks(sbi, inode, false)) 2246 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2247 user_block_count = sbi->user_block_count; 2248 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2249 user_block_count -= sbi->unusable_block_count; 2250 2251 if (unlikely(valid_block_count > user_block_count)) { 2252 spin_unlock(&sbi->stat_lock); 2253 goto enospc; 2254 } 2255 2256 valid_node_count = sbi->total_valid_node_count + 1; 2257 if (unlikely(valid_node_count > sbi->total_node_count)) { 2258 spin_unlock(&sbi->stat_lock); 2259 goto enospc; 2260 } 2261 2262 sbi->total_valid_node_count++; 2263 sbi->total_valid_block_count++; 2264 spin_unlock(&sbi->stat_lock); 2265 2266 if (inode) { 2267 if (is_inode) 2268 f2fs_mark_inode_dirty_sync(inode, true); 2269 else 2270 f2fs_i_blocks_write(inode, 1, true, true); 2271 } 2272 2273 percpu_counter_inc(&sbi->alloc_valid_block_count); 2274 return 0; 2275 2276 enospc: 2277 if (is_inode) { 2278 if (inode) 2279 dquot_free_inode(inode); 2280 } else { 2281 dquot_release_reservation_block(inode, 1); 2282 } 2283 return -ENOSPC; 2284 } 2285 2286 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2287 struct inode *inode, bool is_inode) 2288 { 2289 spin_lock(&sbi->stat_lock); 2290 2291 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2292 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2293 2294 sbi->total_valid_node_count--; 2295 sbi->total_valid_block_count--; 2296 if (sbi->reserved_blocks && 2297 sbi->current_reserved_blocks < sbi->reserved_blocks) 2298 sbi->current_reserved_blocks++; 2299 2300 spin_unlock(&sbi->stat_lock); 2301 2302 if (is_inode) { 2303 dquot_free_inode(inode); 2304 } else { 2305 if (unlikely(inode->i_blocks == 0)) { 2306 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2307 inode->i_ino, 2308 (unsigned long long)inode->i_blocks); 2309 set_sbi_flag(sbi, SBI_NEED_FSCK); 2310 return; 2311 } 2312 f2fs_i_blocks_write(inode, 1, false, true); 2313 } 2314 } 2315 2316 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2317 { 2318 return sbi->total_valid_node_count; 2319 } 2320 2321 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2322 { 2323 percpu_counter_inc(&sbi->total_valid_inode_count); 2324 } 2325 2326 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2327 { 2328 percpu_counter_dec(&sbi->total_valid_inode_count); 2329 } 2330 2331 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2332 { 2333 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2334 } 2335 2336 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2337 pgoff_t index, bool for_write) 2338 { 2339 struct page *page; 2340 2341 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2342 if (!for_write) 2343 page = find_get_page_flags(mapping, index, 2344 FGP_LOCK | FGP_ACCESSED); 2345 else 2346 page = find_lock_page(mapping, index); 2347 if (page) 2348 return page; 2349 2350 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2351 f2fs_show_injection_info(F2FS_M_SB(mapping), 2352 FAULT_PAGE_ALLOC); 2353 return NULL; 2354 } 2355 } 2356 2357 if (!for_write) 2358 return grab_cache_page(mapping, index); 2359 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2360 } 2361 2362 static inline struct page *f2fs_pagecache_get_page( 2363 struct address_space *mapping, pgoff_t index, 2364 int fgp_flags, gfp_t gfp_mask) 2365 { 2366 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2367 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2368 return NULL; 2369 } 2370 2371 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2372 } 2373 2374 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2375 { 2376 char *src_kaddr = kmap(src); 2377 char *dst_kaddr = kmap(dst); 2378 2379 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2380 kunmap(dst); 2381 kunmap(src); 2382 } 2383 2384 static inline void f2fs_put_page(struct page *page, int unlock) 2385 { 2386 if (!page) 2387 return; 2388 2389 if (unlock) { 2390 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2391 unlock_page(page); 2392 } 2393 put_page(page); 2394 } 2395 2396 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2397 { 2398 if (dn->node_page) 2399 f2fs_put_page(dn->node_page, 1); 2400 if (dn->inode_page && dn->node_page != dn->inode_page) 2401 f2fs_put_page(dn->inode_page, 0); 2402 dn->node_page = NULL; 2403 dn->inode_page = NULL; 2404 } 2405 2406 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2407 size_t size) 2408 { 2409 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2410 } 2411 2412 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2413 gfp_t flags) 2414 { 2415 void *entry; 2416 2417 entry = kmem_cache_alloc(cachep, flags); 2418 if (!entry) 2419 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2420 return entry; 2421 } 2422 2423 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2424 { 2425 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2426 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2427 get_pages(sbi, F2FS_WB_CP_DATA) || 2428 get_pages(sbi, F2FS_DIO_READ) || 2429 get_pages(sbi, F2FS_DIO_WRITE)) 2430 return true; 2431 2432 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2433 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2434 return true; 2435 2436 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2437 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2438 return true; 2439 return false; 2440 } 2441 2442 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2443 { 2444 if (sbi->gc_mode == GC_URGENT_HIGH) 2445 return true; 2446 2447 if (is_inflight_io(sbi, type)) 2448 return false; 2449 2450 if (sbi->gc_mode == GC_URGENT_LOW && 2451 (type == DISCARD_TIME || type == GC_TIME)) 2452 return true; 2453 2454 return f2fs_time_over(sbi, type); 2455 } 2456 2457 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2458 unsigned long index, void *item) 2459 { 2460 while (radix_tree_insert(root, index, item)) 2461 cond_resched(); 2462 } 2463 2464 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2465 2466 static inline bool IS_INODE(struct page *page) 2467 { 2468 struct f2fs_node *p = F2FS_NODE(page); 2469 2470 return RAW_IS_INODE(p); 2471 } 2472 2473 static inline int offset_in_addr(struct f2fs_inode *i) 2474 { 2475 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2476 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2477 } 2478 2479 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2480 { 2481 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2482 } 2483 2484 static inline int f2fs_has_extra_attr(struct inode *inode); 2485 static inline block_t data_blkaddr(struct inode *inode, 2486 struct page *node_page, unsigned int offset) 2487 { 2488 struct f2fs_node *raw_node; 2489 __le32 *addr_array; 2490 int base = 0; 2491 bool is_inode = IS_INODE(node_page); 2492 2493 raw_node = F2FS_NODE(node_page); 2494 2495 if (is_inode) { 2496 if (!inode) 2497 /* from GC path only */ 2498 base = offset_in_addr(&raw_node->i); 2499 else if (f2fs_has_extra_attr(inode)) 2500 base = get_extra_isize(inode); 2501 } 2502 2503 addr_array = blkaddr_in_node(raw_node); 2504 return le32_to_cpu(addr_array[base + offset]); 2505 } 2506 2507 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2508 { 2509 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2510 } 2511 2512 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2513 { 2514 int mask; 2515 2516 addr += (nr >> 3); 2517 mask = 1 << (7 - (nr & 0x07)); 2518 return mask & *addr; 2519 } 2520 2521 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2522 { 2523 int mask; 2524 2525 addr += (nr >> 3); 2526 mask = 1 << (7 - (nr & 0x07)); 2527 *addr |= mask; 2528 } 2529 2530 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2531 { 2532 int mask; 2533 2534 addr += (nr >> 3); 2535 mask = 1 << (7 - (nr & 0x07)); 2536 *addr &= ~mask; 2537 } 2538 2539 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2540 { 2541 int mask; 2542 int ret; 2543 2544 addr += (nr >> 3); 2545 mask = 1 << (7 - (nr & 0x07)); 2546 ret = mask & *addr; 2547 *addr |= mask; 2548 return ret; 2549 } 2550 2551 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2552 { 2553 int mask; 2554 int ret; 2555 2556 addr += (nr >> 3); 2557 mask = 1 << (7 - (nr & 0x07)); 2558 ret = mask & *addr; 2559 *addr &= ~mask; 2560 return ret; 2561 } 2562 2563 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2564 { 2565 int mask; 2566 2567 addr += (nr >> 3); 2568 mask = 1 << (7 - (nr & 0x07)); 2569 *addr ^= mask; 2570 } 2571 2572 /* 2573 * On-disk inode flags (f2fs_inode::i_flags) 2574 */ 2575 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2576 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2577 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2578 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2579 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2580 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2581 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2582 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2583 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2584 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2585 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2586 2587 /* Flags that should be inherited by new inodes from their parent. */ 2588 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2589 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2590 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2591 2592 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2593 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2594 F2FS_CASEFOLD_FL)) 2595 2596 /* Flags that are appropriate for non-directories/regular files. */ 2597 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2598 2599 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2600 { 2601 if (S_ISDIR(mode)) 2602 return flags; 2603 else if (S_ISREG(mode)) 2604 return flags & F2FS_REG_FLMASK; 2605 else 2606 return flags & F2FS_OTHER_FLMASK; 2607 } 2608 2609 static inline void __mark_inode_dirty_flag(struct inode *inode, 2610 int flag, bool set) 2611 { 2612 switch (flag) { 2613 case FI_INLINE_XATTR: 2614 case FI_INLINE_DATA: 2615 case FI_INLINE_DENTRY: 2616 case FI_NEW_INODE: 2617 if (set) 2618 return; 2619 fallthrough; 2620 case FI_DATA_EXIST: 2621 case FI_INLINE_DOTS: 2622 case FI_PIN_FILE: 2623 f2fs_mark_inode_dirty_sync(inode, true); 2624 } 2625 } 2626 2627 static inline void set_inode_flag(struct inode *inode, int flag) 2628 { 2629 set_bit(flag, F2FS_I(inode)->flags); 2630 __mark_inode_dirty_flag(inode, flag, true); 2631 } 2632 2633 static inline int is_inode_flag_set(struct inode *inode, int flag) 2634 { 2635 return test_bit(flag, F2FS_I(inode)->flags); 2636 } 2637 2638 static inline void clear_inode_flag(struct inode *inode, int flag) 2639 { 2640 clear_bit(flag, F2FS_I(inode)->flags); 2641 __mark_inode_dirty_flag(inode, flag, false); 2642 } 2643 2644 static inline bool f2fs_verity_in_progress(struct inode *inode) 2645 { 2646 return IS_ENABLED(CONFIG_FS_VERITY) && 2647 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2648 } 2649 2650 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2651 { 2652 F2FS_I(inode)->i_acl_mode = mode; 2653 set_inode_flag(inode, FI_ACL_MODE); 2654 f2fs_mark_inode_dirty_sync(inode, false); 2655 } 2656 2657 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2658 { 2659 if (inc) 2660 inc_nlink(inode); 2661 else 2662 drop_nlink(inode); 2663 f2fs_mark_inode_dirty_sync(inode, true); 2664 } 2665 2666 static inline void f2fs_i_blocks_write(struct inode *inode, 2667 block_t diff, bool add, bool claim) 2668 { 2669 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2670 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2671 2672 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2673 if (add) { 2674 if (claim) 2675 dquot_claim_block(inode, diff); 2676 else 2677 dquot_alloc_block_nofail(inode, diff); 2678 } else { 2679 dquot_free_block(inode, diff); 2680 } 2681 2682 f2fs_mark_inode_dirty_sync(inode, true); 2683 if (clean || recover) 2684 set_inode_flag(inode, FI_AUTO_RECOVER); 2685 } 2686 2687 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2688 { 2689 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2690 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2691 2692 if (i_size_read(inode) == i_size) 2693 return; 2694 2695 i_size_write(inode, i_size); 2696 f2fs_mark_inode_dirty_sync(inode, true); 2697 if (clean || recover) 2698 set_inode_flag(inode, FI_AUTO_RECOVER); 2699 } 2700 2701 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2702 { 2703 F2FS_I(inode)->i_current_depth = depth; 2704 f2fs_mark_inode_dirty_sync(inode, true); 2705 } 2706 2707 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2708 unsigned int count) 2709 { 2710 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2711 f2fs_mark_inode_dirty_sync(inode, true); 2712 } 2713 2714 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2715 { 2716 F2FS_I(inode)->i_xattr_nid = xnid; 2717 f2fs_mark_inode_dirty_sync(inode, true); 2718 } 2719 2720 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2721 { 2722 F2FS_I(inode)->i_pino = pino; 2723 f2fs_mark_inode_dirty_sync(inode, true); 2724 } 2725 2726 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2727 { 2728 struct f2fs_inode_info *fi = F2FS_I(inode); 2729 2730 if (ri->i_inline & F2FS_INLINE_XATTR) 2731 set_bit(FI_INLINE_XATTR, fi->flags); 2732 if (ri->i_inline & F2FS_INLINE_DATA) 2733 set_bit(FI_INLINE_DATA, fi->flags); 2734 if (ri->i_inline & F2FS_INLINE_DENTRY) 2735 set_bit(FI_INLINE_DENTRY, fi->flags); 2736 if (ri->i_inline & F2FS_DATA_EXIST) 2737 set_bit(FI_DATA_EXIST, fi->flags); 2738 if (ri->i_inline & F2FS_INLINE_DOTS) 2739 set_bit(FI_INLINE_DOTS, fi->flags); 2740 if (ri->i_inline & F2FS_EXTRA_ATTR) 2741 set_bit(FI_EXTRA_ATTR, fi->flags); 2742 if (ri->i_inline & F2FS_PIN_FILE) 2743 set_bit(FI_PIN_FILE, fi->flags); 2744 } 2745 2746 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2747 { 2748 ri->i_inline = 0; 2749 2750 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2751 ri->i_inline |= F2FS_INLINE_XATTR; 2752 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2753 ri->i_inline |= F2FS_INLINE_DATA; 2754 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2755 ri->i_inline |= F2FS_INLINE_DENTRY; 2756 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2757 ri->i_inline |= F2FS_DATA_EXIST; 2758 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2759 ri->i_inline |= F2FS_INLINE_DOTS; 2760 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2761 ri->i_inline |= F2FS_EXTRA_ATTR; 2762 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2763 ri->i_inline |= F2FS_PIN_FILE; 2764 } 2765 2766 static inline int f2fs_has_extra_attr(struct inode *inode) 2767 { 2768 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2769 } 2770 2771 static inline int f2fs_has_inline_xattr(struct inode *inode) 2772 { 2773 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2774 } 2775 2776 static inline int f2fs_compressed_file(struct inode *inode) 2777 { 2778 return S_ISREG(inode->i_mode) && 2779 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2780 } 2781 2782 static inline bool f2fs_need_compress_data(struct inode *inode) 2783 { 2784 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2785 2786 if (!f2fs_compressed_file(inode)) 2787 return false; 2788 2789 if (compress_mode == COMPR_MODE_FS) 2790 return true; 2791 else if (compress_mode == COMPR_MODE_USER && 2792 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2793 return true; 2794 2795 return false; 2796 } 2797 2798 static inline unsigned int addrs_per_inode(struct inode *inode) 2799 { 2800 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2801 get_inline_xattr_addrs(inode); 2802 2803 if (!f2fs_compressed_file(inode)) 2804 return addrs; 2805 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2806 } 2807 2808 static inline unsigned int addrs_per_block(struct inode *inode) 2809 { 2810 if (!f2fs_compressed_file(inode)) 2811 return DEF_ADDRS_PER_BLOCK; 2812 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2813 } 2814 2815 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2816 { 2817 struct f2fs_inode *ri = F2FS_INODE(page); 2818 2819 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2820 get_inline_xattr_addrs(inode)]); 2821 } 2822 2823 static inline int inline_xattr_size(struct inode *inode) 2824 { 2825 if (f2fs_has_inline_xattr(inode)) 2826 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2827 return 0; 2828 } 2829 2830 static inline int f2fs_has_inline_data(struct inode *inode) 2831 { 2832 return is_inode_flag_set(inode, FI_INLINE_DATA); 2833 } 2834 2835 static inline int f2fs_exist_data(struct inode *inode) 2836 { 2837 return is_inode_flag_set(inode, FI_DATA_EXIST); 2838 } 2839 2840 static inline int f2fs_has_inline_dots(struct inode *inode) 2841 { 2842 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2843 } 2844 2845 static inline int f2fs_is_mmap_file(struct inode *inode) 2846 { 2847 return is_inode_flag_set(inode, FI_MMAP_FILE); 2848 } 2849 2850 static inline bool f2fs_is_pinned_file(struct inode *inode) 2851 { 2852 return is_inode_flag_set(inode, FI_PIN_FILE); 2853 } 2854 2855 static inline bool f2fs_is_atomic_file(struct inode *inode) 2856 { 2857 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2858 } 2859 2860 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2861 { 2862 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2863 } 2864 2865 static inline bool f2fs_is_volatile_file(struct inode *inode) 2866 { 2867 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2868 } 2869 2870 static inline bool f2fs_is_first_block_written(struct inode *inode) 2871 { 2872 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2873 } 2874 2875 static inline bool f2fs_is_drop_cache(struct inode *inode) 2876 { 2877 return is_inode_flag_set(inode, FI_DROP_CACHE); 2878 } 2879 2880 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2881 { 2882 struct f2fs_inode *ri = F2FS_INODE(page); 2883 int extra_size = get_extra_isize(inode); 2884 2885 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2886 } 2887 2888 static inline int f2fs_has_inline_dentry(struct inode *inode) 2889 { 2890 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2891 } 2892 2893 static inline int is_file(struct inode *inode, int type) 2894 { 2895 return F2FS_I(inode)->i_advise & type; 2896 } 2897 2898 static inline void set_file(struct inode *inode, int type) 2899 { 2900 F2FS_I(inode)->i_advise |= type; 2901 f2fs_mark_inode_dirty_sync(inode, true); 2902 } 2903 2904 static inline void clear_file(struct inode *inode, int type) 2905 { 2906 F2FS_I(inode)->i_advise &= ~type; 2907 f2fs_mark_inode_dirty_sync(inode, true); 2908 } 2909 2910 static inline bool f2fs_is_time_consistent(struct inode *inode) 2911 { 2912 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2913 return false; 2914 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2915 return false; 2916 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2917 return false; 2918 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2919 &F2FS_I(inode)->i_crtime)) 2920 return false; 2921 return true; 2922 } 2923 2924 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2925 { 2926 bool ret; 2927 2928 if (dsync) { 2929 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2930 2931 spin_lock(&sbi->inode_lock[DIRTY_META]); 2932 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2933 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2934 return ret; 2935 } 2936 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2937 file_keep_isize(inode) || 2938 i_size_read(inode) & ~PAGE_MASK) 2939 return false; 2940 2941 if (!f2fs_is_time_consistent(inode)) 2942 return false; 2943 2944 spin_lock(&F2FS_I(inode)->i_size_lock); 2945 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2946 spin_unlock(&F2FS_I(inode)->i_size_lock); 2947 2948 return ret; 2949 } 2950 2951 static inline bool f2fs_readonly(struct super_block *sb) 2952 { 2953 return sb_rdonly(sb); 2954 } 2955 2956 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2957 { 2958 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2959 } 2960 2961 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2962 { 2963 if (len == 1 && name[0] == '.') 2964 return true; 2965 2966 if (len == 2 && name[0] == '.' && name[1] == '.') 2967 return true; 2968 2969 return false; 2970 } 2971 2972 static inline bool f2fs_may_extent_tree(struct inode *inode) 2973 { 2974 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2975 2976 if (!test_opt(sbi, EXTENT_CACHE) || 2977 is_inode_flag_set(inode, FI_NO_EXTENT) || 2978 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2979 return false; 2980 2981 /* 2982 * for recovered files during mount do not create extents 2983 * if shrinker is not registered. 2984 */ 2985 if (list_empty(&sbi->s_list)) 2986 return false; 2987 2988 return S_ISREG(inode->i_mode); 2989 } 2990 2991 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2992 size_t size, gfp_t flags) 2993 { 2994 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2995 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2996 return NULL; 2997 } 2998 2999 return kmalloc(size, flags); 3000 } 3001 3002 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3003 size_t size, gfp_t flags) 3004 { 3005 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3006 } 3007 3008 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3009 size_t size, gfp_t flags) 3010 { 3011 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3012 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3013 return NULL; 3014 } 3015 3016 return kvmalloc(size, flags); 3017 } 3018 3019 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3020 size_t size, gfp_t flags) 3021 { 3022 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3023 } 3024 3025 static inline int get_extra_isize(struct inode *inode) 3026 { 3027 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3028 } 3029 3030 static inline int get_inline_xattr_addrs(struct inode *inode) 3031 { 3032 return F2FS_I(inode)->i_inline_xattr_size; 3033 } 3034 3035 #define f2fs_get_inode_mode(i) \ 3036 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3037 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3038 3039 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3040 (offsetof(struct f2fs_inode, i_extra_end) - \ 3041 offsetof(struct f2fs_inode, i_extra_isize)) \ 3042 3043 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3044 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3045 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3046 sizeof((f2fs_inode)->field)) \ 3047 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3048 3049 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3050 #define MIN_IOSTAT_PERIOD_MS 100 3051 /* maximum period of iostat tracing is 1 day */ 3052 #define MAX_IOSTAT_PERIOD_MS 8640000 3053 3054 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3055 { 3056 int i; 3057 3058 spin_lock(&sbi->iostat_lock); 3059 for (i = 0; i < NR_IO_TYPE; i++) { 3060 sbi->rw_iostat[i] = 0; 3061 sbi->prev_rw_iostat[i] = 0; 3062 } 3063 spin_unlock(&sbi->iostat_lock); 3064 } 3065 3066 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3067 3068 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3069 enum iostat_type type, unsigned long long io_bytes) 3070 { 3071 if (!sbi->iostat_enable) 3072 return; 3073 spin_lock(&sbi->iostat_lock); 3074 sbi->rw_iostat[type] += io_bytes; 3075 3076 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3077 sbi->rw_iostat[APP_BUFFERED_IO] = 3078 sbi->rw_iostat[APP_WRITE_IO] - 3079 sbi->rw_iostat[APP_DIRECT_IO]; 3080 3081 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3082 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3083 sbi->rw_iostat[APP_READ_IO] - 3084 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3085 spin_unlock(&sbi->iostat_lock); 3086 3087 f2fs_record_iostat(sbi); 3088 } 3089 3090 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3091 3092 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3093 3094 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3095 block_t blkaddr, int type); 3096 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3097 block_t blkaddr, int type) 3098 { 3099 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3100 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3101 blkaddr, type); 3102 f2fs_bug_on(sbi, 1); 3103 } 3104 } 3105 3106 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3107 { 3108 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3109 blkaddr == COMPRESS_ADDR) 3110 return false; 3111 return true; 3112 } 3113 3114 static inline void f2fs_set_page_private(struct page *page, 3115 unsigned long data) 3116 { 3117 if (PagePrivate(page)) 3118 return; 3119 3120 attach_page_private(page, (void *)data); 3121 } 3122 3123 static inline void f2fs_clear_page_private(struct page *page) 3124 { 3125 detach_page_private(page); 3126 } 3127 3128 /* 3129 * file.c 3130 */ 3131 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3132 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3133 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3134 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3135 int f2fs_truncate(struct inode *inode); 3136 int f2fs_getattr(const struct path *path, struct kstat *stat, 3137 u32 request_mask, unsigned int flags); 3138 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3139 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3140 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3141 int f2fs_precache_extents(struct inode *inode); 3142 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3143 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3144 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3145 int f2fs_pin_file_control(struct inode *inode, bool inc); 3146 3147 /* 3148 * inode.c 3149 */ 3150 void f2fs_set_inode_flags(struct inode *inode); 3151 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3152 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3153 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3154 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3155 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3156 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3157 void f2fs_update_inode_page(struct inode *inode); 3158 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3159 void f2fs_evict_inode(struct inode *inode); 3160 void f2fs_handle_failed_inode(struct inode *inode); 3161 3162 /* 3163 * namei.c 3164 */ 3165 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3166 bool hot, bool set); 3167 struct dentry *f2fs_get_parent(struct dentry *child); 3168 3169 /* 3170 * dir.c 3171 */ 3172 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3173 int f2fs_init_casefolded_name(const struct inode *dir, 3174 struct f2fs_filename *fname); 3175 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3176 int lookup, struct f2fs_filename *fname); 3177 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3178 struct f2fs_filename *fname); 3179 void f2fs_free_filename(struct f2fs_filename *fname); 3180 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3181 const struct f2fs_filename *fname, int *max_slots); 3182 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3183 unsigned int start_pos, struct fscrypt_str *fstr); 3184 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3185 struct f2fs_dentry_ptr *d); 3186 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3187 const struct f2fs_filename *fname, struct page *dpage); 3188 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3189 unsigned int current_depth); 3190 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3191 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3192 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3193 const struct f2fs_filename *fname, 3194 struct page **res_page); 3195 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3196 const struct qstr *child, struct page **res_page); 3197 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3198 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3199 struct page **page); 3200 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3201 struct page *page, struct inode *inode); 3202 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3203 const struct f2fs_filename *fname); 3204 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3205 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3206 unsigned int bit_pos); 3207 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3208 struct inode *inode, nid_t ino, umode_t mode); 3209 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3210 struct inode *inode, nid_t ino, umode_t mode); 3211 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3212 struct inode *inode, nid_t ino, umode_t mode); 3213 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3214 struct inode *dir, struct inode *inode); 3215 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3216 bool f2fs_empty_dir(struct inode *dir); 3217 3218 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3219 { 3220 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3221 inode, inode->i_ino, inode->i_mode); 3222 } 3223 3224 /* 3225 * super.c 3226 */ 3227 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3228 void f2fs_inode_synced(struct inode *inode); 3229 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3230 int f2fs_quota_sync(struct super_block *sb, int type); 3231 void f2fs_quota_off_umount(struct super_block *sb); 3232 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3233 int f2fs_sync_fs(struct super_block *sb, int sync); 3234 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3235 3236 /* 3237 * hash.c 3238 */ 3239 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3240 3241 /* 3242 * node.c 3243 */ 3244 struct dnode_of_data; 3245 struct node_info; 3246 3247 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3248 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3249 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3250 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3251 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3252 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3253 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3254 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3255 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3256 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3257 struct node_info *ni); 3258 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3259 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3260 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3261 int f2fs_truncate_xattr_node(struct inode *inode); 3262 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3263 unsigned int seq_id); 3264 int f2fs_remove_inode_page(struct inode *inode); 3265 struct page *f2fs_new_inode_page(struct inode *inode); 3266 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3267 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3268 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3269 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3270 int f2fs_move_node_page(struct page *node_page, int gc_type); 3271 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3272 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3273 struct writeback_control *wbc, bool atomic, 3274 unsigned int *seq_id); 3275 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3276 struct writeback_control *wbc, 3277 bool do_balance, enum iostat_type io_type); 3278 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3279 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3280 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3281 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3282 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3283 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3284 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3285 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3286 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3287 unsigned int segno, struct f2fs_summary_block *sum); 3288 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3289 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3290 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3291 int __init f2fs_create_node_manager_caches(void); 3292 void f2fs_destroy_node_manager_caches(void); 3293 3294 /* 3295 * segment.c 3296 */ 3297 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3298 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3299 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3300 void f2fs_drop_inmem_pages(struct inode *inode); 3301 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3302 int f2fs_commit_inmem_pages(struct inode *inode); 3303 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3304 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3305 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3306 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3307 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3308 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3309 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3310 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3311 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3312 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3313 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3314 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3315 struct cp_control *cpc); 3316 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3317 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3318 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3319 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3320 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3321 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3322 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3323 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3324 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3325 unsigned int *newseg, bool new_sec, int dir); 3326 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3327 unsigned int start, unsigned int end); 3328 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3329 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3330 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3331 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3332 struct cp_control *cpc); 3333 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3334 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3335 block_t blk_addr); 3336 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3337 enum iostat_type io_type); 3338 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3339 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3340 struct f2fs_io_info *fio); 3341 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3342 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3343 block_t old_blkaddr, block_t new_blkaddr, 3344 bool recover_curseg, bool recover_newaddr, 3345 bool from_gc); 3346 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3347 block_t old_addr, block_t new_addr, 3348 unsigned char version, bool recover_curseg, 3349 bool recover_newaddr); 3350 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3351 block_t old_blkaddr, block_t *new_blkaddr, 3352 struct f2fs_summary *sum, int type, 3353 struct f2fs_io_info *fio); 3354 void f2fs_wait_on_page_writeback(struct page *page, 3355 enum page_type type, bool ordered, bool locked); 3356 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3357 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3358 block_t len); 3359 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3360 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3361 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3362 unsigned int val, int alloc); 3363 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3364 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3365 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3366 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3367 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3368 int __init f2fs_create_segment_manager_caches(void); 3369 void f2fs_destroy_segment_manager_caches(void); 3370 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3371 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3372 enum page_type type, enum temp_type temp); 3373 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3374 unsigned int segno); 3375 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3376 unsigned int segno); 3377 3378 /* 3379 * checkpoint.c 3380 */ 3381 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3382 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3383 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3384 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3385 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3386 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3387 block_t blkaddr, int type); 3388 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3389 int type, bool sync); 3390 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3391 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3392 long nr_to_write, enum iostat_type io_type); 3393 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3394 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3395 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3396 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3397 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3398 unsigned int devidx, int type); 3399 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3400 unsigned int devidx, int type); 3401 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3402 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3403 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3404 void f2fs_add_orphan_inode(struct inode *inode); 3405 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3406 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3407 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3408 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3409 void f2fs_remove_dirty_inode(struct inode *inode); 3410 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3411 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3412 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3413 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3414 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3415 int __init f2fs_create_checkpoint_caches(void); 3416 void f2fs_destroy_checkpoint_caches(void); 3417 3418 /* 3419 * data.c 3420 */ 3421 int __init f2fs_init_bioset(void); 3422 void f2fs_destroy_bioset(void); 3423 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3424 int f2fs_init_bio_entry_cache(void); 3425 void f2fs_destroy_bio_entry_cache(void); 3426 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3427 struct bio *bio, enum page_type type); 3428 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3429 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3430 struct inode *inode, struct page *page, 3431 nid_t ino, enum page_type type); 3432 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3433 struct bio **bio, struct page *page); 3434 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3435 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3436 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3437 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3438 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3439 block_t blk_addr, struct bio *bio); 3440 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3441 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3442 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3443 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3444 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3445 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3446 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3447 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3448 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3449 int op_flags, bool for_write); 3450 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3451 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3452 bool for_write); 3453 struct page *f2fs_get_new_data_page(struct inode *inode, 3454 struct page *ipage, pgoff_t index, bool new_i_size); 3455 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3456 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3457 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3458 int create, int flag); 3459 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3460 u64 start, u64 len); 3461 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3462 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3463 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3464 int f2fs_write_single_data_page(struct page *page, int *submitted, 3465 struct bio **bio, sector_t *last_block, 3466 struct writeback_control *wbc, 3467 enum iostat_type io_type, 3468 int compr_blocks); 3469 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3470 unsigned int length); 3471 int f2fs_release_page(struct page *page, gfp_t wait); 3472 #ifdef CONFIG_MIGRATION 3473 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3474 struct page *page, enum migrate_mode mode); 3475 #endif 3476 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3477 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3478 int f2fs_init_post_read_processing(void); 3479 void f2fs_destroy_post_read_processing(void); 3480 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3481 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3482 3483 /* 3484 * gc.c 3485 */ 3486 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3487 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3488 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3489 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3490 unsigned int segno); 3491 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3492 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3493 int __init f2fs_create_garbage_collection_cache(void); 3494 void f2fs_destroy_garbage_collection_cache(void); 3495 3496 /* 3497 * recovery.c 3498 */ 3499 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3500 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3501 3502 /* 3503 * debug.c 3504 */ 3505 #ifdef CONFIG_F2FS_STAT_FS 3506 struct f2fs_stat_info { 3507 struct list_head stat_list; 3508 struct f2fs_sb_info *sbi; 3509 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3510 int main_area_segs, main_area_sections, main_area_zones; 3511 unsigned long long hit_largest, hit_cached, hit_rbtree; 3512 unsigned long long hit_total, total_ext; 3513 int ext_tree, zombie_tree, ext_node; 3514 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3515 int ndirty_data, ndirty_qdata; 3516 int inmem_pages; 3517 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3518 int nats, dirty_nats, sits, dirty_sits; 3519 int free_nids, avail_nids, alloc_nids; 3520 int total_count, utilization; 3521 int bg_gc, nr_wb_cp_data, nr_wb_data; 3522 int nr_rd_data, nr_rd_node, nr_rd_meta; 3523 int nr_dio_read, nr_dio_write; 3524 unsigned int io_skip_bggc, other_skip_bggc; 3525 int nr_flushing, nr_flushed, flush_list_empty; 3526 int nr_discarding, nr_discarded; 3527 int nr_discard_cmd; 3528 unsigned int undiscard_blks; 3529 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3530 int compr_inode; 3531 unsigned long long compr_blocks; 3532 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3533 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3534 unsigned int bimodal, avg_vblocks; 3535 int util_free, util_valid, util_invalid; 3536 int rsvd_segs, overp_segs; 3537 int dirty_count, node_pages, meta_pages; 3538 int prefree_count, call_count, cp_count, bg_cp_count; 3539 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3540 int bg_node_segs, bg_data_segs; 3541 int tot_blks, data_blks, node_blks; 3542 int bg_data_blks, bg_node_blks; 3543 unsigned long long skipped_atomic_files[2]; 3544 int curseg[NR_CURSEG_TYPE]; 3545 int cursec[NR_CURSEG_TYPE]; 3546 int curzone[NR_CURSEG_TYPE]; 3547 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3548 unsigned int full_seg[NR_CURSEG_TYPE]; 3549 unsigned int valid_blks[NR_CURSEG_TYPE]; 3550 3551 unsigned int meta_count[META_MAX]; 3552 unsigned int segment_count[2]; 3553 unsigned int block_count[2]; 3554 unsigned int inplace_count; 3555 unsigned long long base_mem, cache_mem, page_mem; 3556 }; 3557 3558 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3559 { 3560 return (struct f2fs_stat_info *)sbi->stat_info; 3561 } 3562 3563 #define stat_inc_cp_count(si) ((si)->cp_count++) 3564 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3565 #define stat_inc_call_count(si) ((si)->call_count++) 3566 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3567 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3568 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3569 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3570 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3571 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3572 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3573 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3574 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3575 #define stat_inc_inline_xattr(inode) \ 3576 do { \ 3577 if (f2fs_has_inline_xattr(inode)) \ 3578 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3579 } while (0) 3580 #define stat_dec_inline_xattr(inode) \ 3581 do { \ 3582 if (f2fs_has_inline_xattr(inode)) \ 3583 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3584 } while (0) 3585 #define stat_inc_inline_inode(inode) \ 3586 do { \ 3587 if (f2fs_has_inline_data(inode)) \ 3588 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3589 } while (0) 3590 #define stat_dec_inline_inode(inode) \ 3591 do { \ 3592 if (f2fs_has_inline_data(inode)) \ 3593 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3594 } while (0) 3595 #define stat_inc_inline_dir(inode) \ 3596 do { \ 3597 if (f2fs_has_inline_dentry(inode)) \ 3598 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3599 } while (0) 3600 #define stat_dec_inline_dir(inode) \ 3601 do { \ 3602 if (f2fs_has_inline_dentry(inode)) \ 3603 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3604 } while (0) 3605 #define stat_inc_compr_inode(inode) \ 3606 do { \ 3607 if (f2fs_compressed_file(inode)) \ 3608 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3609 } while (0) 3610 #define stat_dec_compr_inode(inode) \ 3611 do { \ 3612 if (f2fs_compressed_file(inode)) \ 3613 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3614 } while (0) 3615 #define stat_add_compr_blocks(inode, blocks) \ 3616 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3617 #define stat_sub_compr_blocks(inode, blocks) \ 3618 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3619 #define stat_inc_meta_count(sbi, blkaddr) \ 3620 do { \ 3621 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3622 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3623 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3624 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3625 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3626 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3627 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3628 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3629 } while (0) 3630 #define stat_inc_seg_type(sbi, curseg) \ 3631 ((sbi)->segment_count[(curseg)->alloc_type]++) 3632 #define stat_inc_block_count(sbi, curseg) \ 3633 ((sbi)->block_count[(curseg)->alloc_type]++) 3634 #define stat_inc_inplace_blocks(sbi) \ 3635 (atomic_inc(&(sbi)->inplace_count)) 3636 #define stat_update_max_atomic_write(inode) \ 3637 do { \ 3638 int cur = F2FS_I_SB(inode)->atomic_files; \ 3639 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3640 if (cur > max) \ 3641 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3642 } while (0) 3643 #define stat_inc_volatile_write(inode) \ 3644 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3645 #define stat_dec_volatile_write(inode) \ 3646 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3647 #define stat_update_max_volatile_write(inode) \ 3648 do { \ 3649 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3650 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3651 if (cur > max) \ 3652 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3653 } while (0) 3654 #define stat_inc_seg_count(sbi, type, gc_type) \ 3655 do { \ 3656 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3657 si->tot_segs++; \ 3658 if ((type) == SUM_TYPE_DATA) { \ 3659 si->data_segs++; \ 3660 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3661 } else { \ 3662 si->node_segs++; \ 3663 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3664 } \ 3665 } while (0) 3666 3667 #define stat_inc_tot_blk_count(si, blks) \ 3668 ((si)->tot_blks += (blks)) 3669 3670 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3671 do { \ 3672 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3673 stat_inc_tot_blk_count(si, blks); \ 3674 si->data_blks += (blks); \ 3675 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3676 } while (0) 3677 3678 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3679 do { \ 3680 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3681 stat_inc_tot_blk_count(si, blks); \ 3682 si->node_blks += (blks); \ 3683 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3684 } while (0) 3685 3686 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3687 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3688 void __init f2fs_create_root_stats(void); 3689 void f2fs_destroy_root_stats(void); 3690 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3691 #else 3692 #define stat_inc_cp_count(si) do { } while (0) 3693 #define stat_inc_bg_cp_count(si) do { } while (0) 3694 #define stat_inc_call_count(si) do { } while (0) 3695 #define stat_inc_bggc_count(si) do { } while (0) 3696 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3697 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3698 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3699 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3700 #define stat_inc_total_hit(sbi) do { } while (0) 3701 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3702 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3703 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3704 #define stat_inc_inline_xattr(inode) do { } while (0) 3705 #define stat_dec_inline_xattr(inode) do { } while (0) 3706 #define stat_inc_inline_inode(inode) do { } while (0) 3707 #define stat_dec_inline_inode(inode) do { } while (0) 3708 #define stat_inc_inline_dir(inode) do { } while (0) 3709 #define stat_dec_inline_dir(inode) do { } while (0) 3710 #define stat_inc_compr_inode(inode) do { } while (0) 3711 #define stat_dec_compr_inode(inode) do { } while (0) 3712 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3713 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3714 #define stat_inc_atomic_write(inode) do { } while (0) 3715 #define stat_dec_atomic_write(inode) do { } while (0) 3716 #define stat_update_max_atomic_write(inode) do { } while (0) 3717 #define stat_inc_volatile_write(inode) do { } while (0) 3718 #define stat_dec_volatile_write(inode) do { } while (0) 3719 #define stat_update_max_volatile_write(inode) do { } while (0) 3720 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3721 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3722 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3723 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3724 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3725 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3726 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3727 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3728 3729 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3730 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3731 static inline void __init f2fs_create_root_stats(void) { } 3732 static inline void f2fs_destroy_root_stats(void) { } 3733 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3734 #endif 3735 3736 extern const struct file_operations f2fs_dir_operations; 3737 extern const struct file_operations f2fs_file_operations; 3738 extern const struct inode_operations f2fs_file_inode_operations; 3739 extern const struct address_space_operations f2fs_dblock_aops; 3740 extern const struct address_space_operations f2fs_node_aops; 3741 extern const struct address_space_operations f2fs_meta_aops; 3742 extern const struct inode_operations f2fs_dir_inode_operations; 3743 extern const struct inode_operations f2fs_symlink_inode_operations; 3744 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3745 extern const struct inode_operations f2fs_special_inode_operations; 3746 extern struct kmem_cache *f2fs_inode_entry_slab; 3747 3748 /* 3749 * inline.c 3750 */ 3751 bool f2fs_may_inline_data(struct inode *inode); 3752 bool f2fs_may_inline_dentry(struct inode *inode); 3753 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3754 void f2fs_truncate_inline_inode(struct inode *inode, 3755 struct page *ipage, u64 from); 3756 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3757 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3758 int f2fs_convert_inline_inode(struct inode *inode); 3759 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3760 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3761 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3762 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3763 const struct f2fs_filename *fname, 3764 struct page **res_page); 3765 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3766 struct page *ipage); 3767 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3768 struct inode *inode, nid_t ino, umode_t mode); 3769 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3770 struct page *page, struct inode *dir, 3771 struct inode *inode); 3772 bool f2fs_empty_inline_dir(struct inode *dir); 3773 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3774 struct fscrypt_str *fstr); 3775 int f2fs_inline_data_fiemap(struct inode *inode, 3776 struct fiemap_extent_info *fieinfo, 3777 __u64 start, __u64 len); 3778 3779 /* 3780 * shrinker.c 3781 */ 3782 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3783 struct shrink_control *sc); 3784 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3785 struct shrink_control *sc); 3786 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3787 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3788 3789 /* 3790 * extent_cache.c 3791 */ 3792 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3793 struct rb_entry *cached_re, unsigned int ofs); 3794 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3795 struct rb_root_cached *root, 3796 struct rb_node **parent, 3797 unsigned long long key, bool *left_most); 3798 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3799 struct rb_root_cached *root, 3800 struct rb_node **parent, 3801 unsigned int ofs, bool *leftmost); 3802 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3803 struct rb_entry *cached_re, unsigned int ofs, 3804 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3805 struct rb_node ***insert_p, struct rb_node **insert_parent, 3806 bool force, bool *leftmost); 3807 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3808 struct rb_root_cached *root, bool check_key); 3809 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3810 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3811 void f2fs_drop_extent_tree(struct inode *inode); 3812 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3813 void f2fs_destroy_extent_tree(struct inode *inode); 3814 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3815 struct extent_info *ei); 3816 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3817 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3818 pgoff_t fofs, block_t blkaddr, unsigned int len); 3819 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3820 int __init f2fs_create_extent_cache(void); 3821 void f2fs_destroy_extent_cache(void); 3822 3823 /* 3824 * sysfs.c 3825 */ 3826 int __init f2fs_init_sysfs(void); 3827 void f2fs_exit_sysfs(void); 3828 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3829 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3830 3831 /* verity.c */ 3832 extern const struct fsverity_operations f2fs_verityops; 3833 3834 /* 3835 * crypto support 3836 */ 3837 static inline bool f2fs_encrypted_file(struct inode *inode) 3838 { 3839 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3840 } 3841 3842 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3843 { 3844 #ifdef CONFIG_FS_ENCRYPTION 3845 file_set_encrypt(inode); 3846 f2fs_set_inode_flags(inode); 3847 #endif 3848 } 3849 3850 /* 3851 * Returns true if the reads of the inode's data need to undergo some 3852 * postprocessing step, like decryption or authenticity verification. 3853 */ 3854 static inline bool f2fs_post_read_required(struct inode *inode) 3855 { 3856 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3857 f2fs_compressed_file(inode); 3858 } 3859 3860 /* 3861 * compress.c 3862 */ 3863 #ifdef CONFIG_F2FS_FS_COMPRESSION 3864 bool f2fs_is_compressed_page(struct page *page); 3865 struct page *f2fs_compress_control_page(struct page *page); 3866 int f2fs_prepare_compress_overwrite(struct inode *inode, 3867 struct page **pagep, pgoff_t index, void **fsdata); 3868 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3869 pgoff_t index, unsigned copied); 3870 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3871 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3872 bool f2fs_is_compress_backend_ready(struct inode *inode); 3873 int f2fs_init_compress_mempool(void); 3874 void f2fs_destroy_compress_mempool(void); 3875 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3876 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3877 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3878 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3879 int f2fs_write_multi_pages(struct compress_ctx *cc, 3880 int *submitted, 3881 struct writeback_control *wbc, 3882 enum iostat_type io_type); 3883 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3884 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3885 unsigned nr_pages, sector_t *last_block_in_bio, 3886 bool is_readahead, bool for_write); 3887 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3888 void f2fs_free_dic(struct decompress_io_ctx *dic); 3889 void f2fs_decompress_end_io(struct page **rpages, 3890 unsigned int cluster_size, bool err, bool verity); 3891 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3892 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3893 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3894 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3895 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3896 int __init f2fs_init_compress_cache(void); 3897 void f2fs_destroy_compress_cache(void); 3898 #else 3899 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3900 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3901 { 3902 if (!f2fs_compressed_file(inode)) 3903 return true; 3904 /* not support compression */ 3905 return false; 3906 } 3907 static inline struct page *f2fs_compress_control_page(struct page *page) 3908 { 3909 WARN_ON_ONCE(1); 3910 return ERR_PTR(-EINVAL); 3911 } 3912 static inline int f2fs_init_compress_mempool(void) { return 0; } 3913 static inline void f2fs_destroy_compress_mempool(void) { } 3914 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3915 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3916 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3917 static inline void f2fs_destroy_compress_cache(void) { } 3918 #endif 3919 3920 static inline void set_compress_context(struct inode *inode) 3921 { 3922 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3923 3924 F2FS_I(inode)->i_compress_algorithm = 3925 F2FS_OPTION(sbi).compress_algorithm; 3926 F2FS_I(inode)->i_log_cluster_size = 3927 F2FS_OPTION(sbi).compress_log_size; 3928 F2FS_I(inode)->i_compress_flag = 3929 F2FS_OPTION(sbi).compress_chksum ? 3930 1 << COMPRESS_CHKSUM : 0; 3931 F2FS_I(inode)->i_cluster_size = 3932 1 << F2FS_I(inode)->i_log_cluster_size; 3933 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3934 set_inode_flag(inode, FI_COMPRESSED_FILE); 3935 stat_inc_compr_inode(inode); 3936 f2fs_mark_inode_dirty_sync(inode, true); 3937 } 3938 3939 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3940 { 3941 struct f2fs_inode_info *fi = F2FS_I(inode); 3942 3943 if (!f2fs_compressed_file(inode)) 3944 return true; 3945 if (S_ISREG(inode->i_mode) && 3946 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3947 return false; 3948 3949 fi->i_flags &= ~F2FS_COMPR_FL; 3950 stat_dec_compr_inode(inode); 3951 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3952 f2fs_mark_inode_dirty_sync(inode, true); 3953 return true; 3954 } 3955 3956 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3957 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3958 { \ 3959 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3960 } 3961 3962 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3963 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3964 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3965 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3966 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3967 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3968 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3969 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3970 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3971 F2FS_FEATURE_FUNCS(verity, VERITY); 3972 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3973 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3974 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3975 3976 #ifdef CONFIG_BLK_DEV_ZONED 3977 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3978 block_t blkaddr) 3979 { 3980 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3981 3982 return test_bit(zno, FDEV(devi).blkz_seq); 3983 } 3984 #endif 3985 3986 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3987 { 3988 return f2fs_sb_has_blkzoned(sbi); 3989 } 3990 3991 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3992 { 3993 return blk_queue_discard(bdev_get_queue(bdev)) || 3994 bdev_is_zoned(bdev); 3995 } 3996 3997 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3998 { 3999 int i; 4000 4001 if (!f2fs_is_multi_device(sbi)) 4002 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4003 4004 for (i = 0; i < sbi->s_ndevs; i++) 4005 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4006 return true; 4007 return false; 4008 } 4009 4010 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4011 { 4012 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4013 f2fs_hw_should_discard(sbi); 4014 } 4015 4016 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4017 { 4018 int i; 4019 4020 if (!f2fs_is_multi_device(sbi)) 4021 return bdev_read_only(sbi->sb->s_bdev); 4022 4023 for (i = 0; i < sbi->s_ndevs; i++) 4024 if (bdev_read_only(FDEV(i).bdev)) 4025 return true; 4026 return false; 4027 } 4028 4029 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4030 { 4031 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4032 } 4033 4034 static inline bool f2fs_may_compress(struct inode *inode) 4035 { 4036 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4037 f2fs_is_atomic_file(inode) || 4038 f2fs_is_volatile_file(inode)) 4039 return false; 4040 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4041 } 4042 4043 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4044 u64 blocks, bool add) 4045 { 4046 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4047 struct f2fs_inode_info *fi = F2FS_I(inode); 4048 4049 /* don't update i_compr_blocks if saved blocks were released */ 4050 if (!add && !atomic_read(&fi->i_compr_blocks)) 4051 return; 4052 4053 if (add) { 4054 atomic_add(diff, &fi->i_compr_blocks); 4055 stat_add_compr_blocks(inode, diff); 4056 } else { 4057 atomic_sub(diff, &fi->i_compr_blocks); 4058 stat_sub_compr_blocks(inode, diff); 4059 } 4060 f2fs_mark_inode_dirty_sync(inode, true); 4061 } 4062 4063 static inline int block_unaligned_IO(struct inode *inode, 4064 struct kiocb *iocb, struct iov_iter *iter) 4065 { 4066 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4067 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4068 loff_t offset = iocb->ki_pos; 4069 unsigned long align = offset | iov_iter_alignment(iter); 4070 4071 return align & blocksize_mask; 4072 } 4073 4074 static inline int allow_outplace_dio(struct inode *inode, 4075 struct kiocb *iocb, struct iov_iter *iter) 4076 { 4077 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4078 int rw = iov_iter_rw(iter); 4079 4080 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4081 !block_unaligned_IO(inode, iocb, iter)); 4082 } 4083 4084 static inline bool f2fs_force_buffered_io(struct inode *inode, 4085 struct kiocb *iocb, struct iov_iter *iter) 4086 { 4087 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4088 int rw = iov_iter_rw(iter); 4089 4090 if (f2fs_post_read_required(inode)) 4091 return true; 4092 if (f2fs_is_multi_device(sbi)) 4093 return true; 4094 /* 4095 * for blkzoned device, fallback direct IO to buffered IO, so 4096 * all IOs can be serialized by log-structured write. 4097 */ 4098 if (f2fs_sb_has_blkzoned(sbi)) 4099 return true; 4100 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4101 if (block_unaligned_IO(inode, iocb, iter)) 4102 return true; 4103 if (F2FS_IO_ALIGNED(sbi)) 4104 return true; 4105 } 4106 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4107 !IS_SWAPFILE(inode)) 4108 return true; 4109 4110 return false; 4111 } 4112 4113 #ifdef CONFIG_F2FS_FAULT_INJECTION 4114 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4115 unsigned int type); 4116 #else 4117 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4118 #endif 4119 4120 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4121 { 4122 #ifdef CONFIG_QUOTA 4123 if (f2fs_sb_has_quota_ino(sbi)) 4124 return true; 4125 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4126 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4127 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4128 return true; 4129 #endif 4130 return false; 4131 } 4132 4133 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4134 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4135 4136 #endif /* _LINUX_F2FS_H */ 4137